Book Changelog
This Machine Learning Systems textbook is constantly evolving. This changelog automatically records all updates and improvements, helping you stay informed about whatβs new and refined.
Changelog
π 2025
June 10
π Frontmatter
βββββSocratiQ: Page links for SocratiQ were updated and documentation for the SocratiQ AI learning companion was addedβββββSocratiQ: The SocratiQ AI feature has been removed and minor grammatical and content errors have been correctedβββββAbout: SocratiQ page links were updated
π Chapters
βββββChapter 3: DL Primer: Resource sections were added to the chapter, the distinction between training and inference was clarified, and several figures were incorporatedβββββChapter 4: DNN Architectures: Resource sections were added to the chapter content, text processing in QMD files was improved, and explanations of deep learning architectures were refinedβββββChapter 5: AI Workflow: This chapter now includes resource sections, benefits from improved text processing in QMD files, features enhanced clarity and consistency, and incorporates new figuresβββββChapter 6: Data Engineering: This chapter now includes a data pipeline overview diagram, utilizes TikZ for diagrams instead of Mermaid, and removes a broken web scraping Colab linkβββββChapter 9: Efficient AI: Resource sections were added to the chapter content, text processing in QMD files was improved, and the compute-optimal scaling frontier was clarifiedβββββChapter 10: Model Optimizations: This update clarifies pruning strategies, incorporates resource sections into the core content, and refines the documentation on model optimization techniquesβββββChapter 13: ML Operations: Resource sections were added to the chapter, MLOps content was updated for clarity and accuracy, and the operations diagram and text were revisedβββββChapter 14: On-Device Learning: The chapter now includes resource sections, the equations for adapter-based adaptation are clarified, and explanations and concepts have been refinedβββββChapter 18: Robust AI: This update includes resource sections, enhanced text processing within QMD files, and a new figure environment for error maskingβββββChapter 15: Security & Privacy: Added a threat mitigation flow diagram, illustrated model theft attack types, and included an ML lifecycle threat diagramβββββChapter 16: Responsible AI: Expanded the Responsible AI chapter with new sections on governance structures, safety and robustness considerations, privacy architectures, fairness constraints, system explainability, deployment contexts, and detailed discussions on privacy, data governance, fairness, and explainability/transparencyβββββChapter 19: AI for Good: Resource sections were added to the chapter, text processing in QMD files was improved, and the AI for Good content was refined for clarityβββββChapter 1: Introduction: The introduction chapter now includes updated content, added footnotes, corrected minor grammatical errors, and refined language for clarityβββββChapter 2: ML Systems: Resource sections were added to the chapter content, the TinyML example callout was removed, and minor grammatical and content errors were correctedβββββChapter 7: AI Frameworks: Resource sections were added to the chapter content, text processing in QMD files was improved, and figure references along with a parallelization typo were fixedβββββChapter 8: AI Training: Resource sections were added to the chapter content, text processing in QMD files was improved, and the explanation of activation checkpointing was clarifiedβββββChapter 11: AI Acceleration: Resource sections were added to the chapter, placement and allocation definitions were clarified, and code block language was corrected for clarityβββββChapter 12: Benchmarking AI: Resource sections were added to the chapter content, text processing in QMD files was improved, and a typo in the benchmarking documentation was fixedβββββChapter 17: Sustainable AI: Resource sections were added to the chapter content, minor grammatical and content errors were corrected, and text processing in QMD files was improved
π§βπ» Labs
βββββLab: Grove Vision Ai V2: The Grove Vision AI V2 lab now includes improved documentation and a new lab exerciseβββββLab: Setup And No Code Apps: A new lab on Grove Vision AI v2 has been added and the existing documentation has been improved with fixes for typos and increased clarityβββββLab: Arduino Image Classification: A new lab demonstrating image classification with Grove Vision AI V2 has been added along with improvements to the Grove Vision AI V2 documentationβββββLab: Arduino Object Detection: A new lab has been added demonstrating object detection using the Grove Vision AI v2 module on an ArduinoβββββLab: Lab Setup: The lab setup instructions were updated, links to required software were added, and minor typos were correctedβββββLab: Labs Overview: The lab file templates were updated and instructions for using Jupyter notebooks were addedβββββLab: XIAO Image Classification: A typo was corrected in the image classification lab instructions
May 14
π Chapters
βββββChapter 14: On-Device Learning: The on-device learning content has been refactored and clarified for improved understanding
May 14
π Chapters
βββββChapter 14: On-Device Learning: The on-device learning content has been refactored and clarified for improved understanding
May 04
π Chapters
βββββChapter 5: AI Workflow: The chapter now includes updated explanations of data preprocessing and a new section on model evaluation metricsβββββChapter 6: Data Engineering: The chapter now includes updated examples for data pipelines and a new section on cloud-based data engineering toolsβββββChapter 10: Model Optimizations: The chapter now includes updated explanations of gradient descent variants and additional examples demonstrating momentum optimization techniquesβββββChapter 11: AI Acceleration: The discussion on hardware acceleration has been refined, the explanation of hardware specialization has been enhanced, and the explanation of AI compute primitives has been clarifiedβββββChapter 13: ML Operations: The chapter now features consolidated TikZ figure styling, expanded MLOps core concepts and case studies, and corrected acronym usageβββββChapter 14: On-Device Learning: This update includes a new Federated Learning section, explores on-device learning strategies for limited data scenarios, and expands the discussion on security considerations in on-device learningβββββChapter 1: Introduction: The introduction has been updated with improved content and clarityβββββChapter 3: DL Primer: The dimension order for W^L was swapped and a script was added to identify any missing referencesβββββChapter 4: DNN Architectures: The chapter now includes a script to identify and address any missing references within the textβββββChapter 7: AI Frameworks: Content from Chapter 6 was integrated and various fixes were implementedβββββChapter 8: AI Training: Minor issues were fixed and label checking was improvedβββββChapter 12: Benchmarking AI: The chapterβs βPurposeβ heading was renamed to βMotivationβ, benchmarking metrics and power measurements were clarified, and a script was added to identify missing referencesβββββChapter 18: Robust AI: Implemented label checking fixes and a script to identify missing referencesβββββChapter 15: Security & Privacy: Minor issues were fixed and label checking was improvedβββββChapter 17: Sustainable AI: Figure styling was consolidated using TikZ and a script was created to identify missing referencesβββββChapter 19: AI for Good: The PlantVillage Nuru footnote accuracy was updated and a script was added to locate any missing referencesβββββChapter 2: ML Systems: The wake-word detection footnote was formatted, a redundant footnote definition was removed, and the grammar in the GDPR/HIPAA footnote was correctedβββββChapter 9: Efficient AI: Bravo fixed several errors in the text of Chapter 9
π§βπ» Labs
βββββLab: Arduino Keyword Spotting: Fixed keyword recognition accuracy issues, updated code for better performance, and added explanations for model training parametersβββββLab: Raspi: Updated Raspberry Pi setup instructions, added code for an image classification task, and improved the troubleshooting guideβββββLab: Raspberry Pi Setup: The Raspberry Pi setup instructions were updated with links to the latest software versions and added troubleshooting tipsβββββLab: Arduino Setup: The Nicla setup instructions were updated for improved clarity and accuracyβββββLab: Arduino Image Classification: The labβs formatting, wording, instructions, documentation, text content, and images have been updated for clarity and accuracyβββββLab: Arduino Object Detection: The lab now includes formatting updates, clearer instructions, and an updated object detection modelβββββLab: Arduino Motion Classification: The code example was corrected and the instructions for connecting the NICLA Vision board were updatedβββββLab: Xiao Esp32S3: The Xiao Esp32S3 lab content has been updatedβββββLab: XIAO Setup: Updated instructions for flashing firmware, added troubleshooting tips, and included links to Seeed documentationβββββLab: XIAO Image Classification: Fixes were made to the LABS part 2_seeed_xiao_esp32s3βββββLab: XIAO Object Detection: The XIAO Object Detection lab has been updated with changes to LABS part 2_seeed_xiao_esp32s3βββββLab: XIAO Keyword Spotting: The XIAO keyword spotting lab code has been updated and accompanied by explanations for the modificationsβββββLab: XIAO Motion Classification: The XIAO Motion Classification lab now includes updated code and explanations for sensor data processingβββββLab: Pi Image Classification: The Raspberry Pi image classification code was updated and accompanied by explanations for the modificationsβββββLab: Pi Object Detection: The object detection lab now includes updated code and explanations for setting up a Raspberry PiβββββLab: Pi Large Language Models: The Raspberry Pi labs section has been updatedβββββLab: Pi Vision Language Models: The VLM lab guide was refactored for improved clarity and the LABS part_raspi was updatedβββββLab: Dsp Spectral Features Block: The LABS part_shared has been updatedβββββLab: Kws Feature Eng: The shared LABS section has been updated
March 26
π Chapters
βββββChapter 13: ML Operations: An exercise reference was removed and another was correctedβββββChapter 18: Robust AI: Package commands were fixed, Helvetica font was added, and line commands were corrected
March 25
π Frontmatter
βββββForeword: The foreword has been updated based on feedback from vale testingβββββAbout: The book underwent vale testing for improvementsβββββAcknowledgements: The contributors list has been updatedβββββSocratiQ: Trailing whitespace was removed and all broken links were fixed
π Chapters
βββββChapter 9: Efficient AI: A new Scaling Laws section was added to the chapter, section headers were updated, and epoch AI figures were includedβββββChapter 13: ML Operations: The MLOps key components section was updated, embedded Operations were replaced with Operational design principles, and references were addedβββββChapter 18: Robust AI: The chapter now includes an expanded attacks section, a more detailed real-world applications section, an introductory paragraph, and a discussion on shifts in adversarial examplesβββββChapter 1: Introduction: The introduction chapter now has consistent footnote naming and removed unnecessary code and packages for a cleaner build processβββββChapter 2: ML Systems: Fixed hyphenation issues, corrected a typo impacting PDF generation, standardized Markdown styles, incorporated footnotes for improved clarity, and addressed inconsistencies in footnote namingβββββChapter 4: DNN Architectures: Hyphenation is now handled globally, colons are consistently formatted, and all broken links and missing figure references have been correctedβββββChapter 5: AI Workflow: A definition was added and references were updatedβββββChapter 6: Data Engineering: The chapter now uses consistent hyphenation, has updated markdown styles, includes a mermaid diagram, and features several text edits for clarityβββββChapter 7: AI Frameworks: A figure illustrating ONNX was added, footnote naming conventions were standardized, and the callout format was correctedβββββChapter 8: AI Training: Footnote naming was standardized, missing figure references were fixed, and section headers were reviewed for consistencyβββββChapter 10: Model Optimizations: The chapter received spelling and stylistic improvements, including added footnotes and corrected figure references, along with a fix for broken linksβββββChapter 11: AI Acceleration: The chapter received spelling corrections, updated Markdown styles for consistency, and improved footnote and figure reference formattingβββββChapter 12: Benchmarking AI: The chapter now has corrected links, updated section headers, and a resolved missing figure referenceβββββChapter 14: On-Device Learning: The chapterβs markdown styling was cleaned up and all broken links were fixedβββββChapter 15: Security & Privacy: Markdown styles were cleaned and all broken links were fixedβββββChapter 16: Responsible AI: The chapterβs markdown formatting has been cleaned up and all broken links have been fixedβββββChapter 17: Sustainable AI: This update includes a plot and discussion on Jevonβs paradox, along with fixes for broken links within the chapterβββββChapter 19: AI for Good: The chapter now has consistent footnote naming, fixed broken links, and improved Markdown styling with the addition of a new definitionβββββChapter 20: Conclusion: Dangling sections were removed and the conclusion underwent a first round of editingβββββChapter: Old Sus Ai: The chapter content has been updated and an outdated file removedβββββChapter 3: DL Primer: Markdown styles were cleaned, a definition was added, footnotes were updated for consistency, and footnote naming conventions were standardizedβββββChapter: Generative Ai: The Generative AI chapter received its initial draft
π§βπ» Labs
βββββLab: Arduino Setup: The Arduino setup instructions now have consistent Markdown styling and all links are functionalβββββLab: Arduino Motion Classification: The labβs Markdown formatting was cleaned up and all broken links were fixedβββββLab: Pi Vision Language Models: The labβs Markdown formatting has been standardized and all broken links have been repairedβββββLab: Arduino Image Classification: Markdown styles were cleaned up in the image classification labβββββLab: Arduino Keyword Spotting: The Markdown styles in the file have been cleaned upβββββLab: XIAO Image Classification: Spelling mistakes have been fixed throughout the labβββββLab: XIAO Keyword Spotting: Spelling errors were fixed and Markdown styles were standardized for consistencyβββββLab: XIAO Motion Classification: The labβs Markdown formatting has been updated for improved readability and style consistencyβββββLab: Raspberry Pi Setup: Spelling mistakes were fixed using codespell checkβββββLab: Pi Object Detection: The object detection labβs Markdown file now has fixed links and improved stylingβββββLab: Pi Large Language Models: Markdown styles were cleaned up throughout the fileβββββLab: Kws Feature Eng: Markdown styles were cleaned up throughout the labβββββLab: Dsp Spectral Features Block: Corrected spelling errors and standardized markdown formatting
π Appendix
βββββPhD Survival Guide: The PhD Survival Guide now has corrected spelling and all broken links have been resolved
March 03
π Frontmatter
βββββSocratiQ: Callout titles are now formatted correctly and QMD file linting has been addressedβββββAbout: Header spacing has been adjusted for consistencyβββββAcknowledgements: The contributors list has been updated and reflected in the readme and acknowledgements file
π Chapters
βββββChapter 10: Model Optimizations: This chapter now includes sections on structured optimization, post-training quantization (PTQ) and quantization aware training (QAT), sparsity visualization techniques, knowledge distillation (KD) figures, and learning transfer hardening (LTH) along with iterative pruning and calibrationβββββChapter 11: AI Acceleration: This update includes a new section on hybrid mapping strategies and a plot comparing model performance to memory bandwidth. The chapter overview and definition have also been updatedβββββChapter 1: Introduction: Fixed callout title formatting and resolved issues with QMD file lintingβββββChapter 3: DL Primer: Formatting and linting improvements were made to the DL Primer chapterβββββChapter 4: DNN Architectures: Callout title formatting and QMD file linting were fixedβββββChapter 5: AI Workflow: Redundant definitions were removed, text errors were corrected, and QMD file linting was fixedβββββChapter 6: Data Engineering: Formatting and code issues were addressed to improve readability and functionalityβββββChapter 7: AI Frameworks: Added model and data parallelism images for distributed execution, removed redundant definitions, and updated the chapter by removing an unnecessary log fileβββββChapter 8: AI Training: Descriptions of single and multi GPU systems were added, redundant definitions were removed, and MD formatting issues were fixedβββββChapter 9: Efficient AI: Redundant definitions were removed, spacing issues were fixed, and QMD file linting was improvedβββββChapter 12: Benchmarking AI: The chapter now uses tikz code for images, includes a new image illustrating datacentric AI, and has corrected a broken reference and citationβββββChapter 13: ML Operations: Definitions were streamlined, callout titles are now correctly formatted, and QMD file linting issues have been resolvedβββββChapter 18: Robust AI: Redundant definitions were removed and various formatting issues in Markdown, callouts, and general text were correctedβββββChapter 15: Security & Privacy: Definitions were removed, markdown formatting was fixed, and callout titles were correctedβββββChapter 17: Sustainable AI: Definitions were removed, callout titles were formatted correctly, and QMD file linting was fixedβββββChapter 19: AI for Good: Definitions were removed, callout titles were formatted correctly, and QMD file linting was fixedβββββChapter 2: ML Systems: Markdown formatting and linting were addressed in the chapterβββββChapter 14: On-Device Learning: Typos were corrected, redundant definitions were removed, and callout titles were formatted correctlyβββββChapter 16: Responsible AI: Callout title formatting and QMD file linting were fixed
π§βπ» Labs
βββββLab: Nicla Vision: The Nicla Vision LABS section has been updatedβββββLab: Arduino Setup: The Arduino setup section for the nicla_vision LABS has been updatedβββββLab: Arduino Image Classification: The Arduino image classification lab has been updated with changes to the arduino/nicla_vision LABS sectionβββββLab: Arduino Object Detection: The Arduino object detection lab has been updated with the latest changes to the arduino/nicla_vision repositoryβββββLab: Arduino Keyword Spotting: The Arduino/Nicla Vision LABS section has been updatedβββββLab: Arduino Motion Classification: The Arduino motion classification lab content was updated and header spacing was improved for consistencyβββββLab: Pi Vision Language Models: The lab now has corrected markdown formatting, improved linting for QMD files, and adjusted header spacingβββββLab: Kws Feature Eng: The lab now has corrected markdown formatting and resolved linting errors in the QMD filesβββββLab: Labs Overview: Linting errors in QMD files were corrected
February 08
π Frontmatter
βββββAcknowledgements: The contributor list has been updated to include all contributors and correct any omissionsβββββSocratiQ: Included a link to the Socratiq research paper
February 07
π Frontmatter
βββββIndex: The precheck now only runs on .qmd and .bib filesβββββAbout: The precheck script now only runs on qmd and bib filesβββββChangelog: The changelog is now automatically generated and formatted within the frontmatterβββββSocratiQ: The SocratiQ precheck now only runs on qmd and bib filesβββββAcknowledgements: The acknowledgements section now includes a list of contributors and the readme has been updated to reflect this
π Chapters
βββββChapter 12: Benchmarking AI: The chapter now includes MLPerf power trends data, a FastML science graph visualization, and updated benchmarking challenges with new plots and graphs to motivate the importance of benchmarkingβββββChapter 1: Introduction: The code now directly uses _quarto.yml and the precheck script is limited to qmd and bib filesβββββChapter 2: ML Systems: The ML Systems chapter content was updated and the precheck script now only runs on .qmd and .bib filesβββββChapter 3: DL Primer: The precheck script now only runs on .qmd and .bib filesβββββChapter 4: DNN Architectures: The precheck script now only runs on qmd and bib filesβββββChapter 5: AI Workflow: The precheck function now only runs on .qmd and .bib filesβββββChapter 6: Data Engineering: The precheck script now only runs on qmd and bib filesβββββChapter 7: AI Frameworks: The code now relies solely on _quarto.yml and the precheck function is limited to qmd and bib filesβββββChapter 8: AI Training: Figure formatting was fixed, a diagram was updated, and an unused library was removed from the codeβββββChapter 9: Efficient AI: The code now uses _quarto.yml instead of an in-use library, a bib reference was corrected, and R code was added for debugging and visualization purposesβββββChapter 10: Model Optimizations: The precheck script now only runs on qmd and bib filesβββββChapter 11: AI Acceleration: The precheck script now only runs on qmd and bib filesβββββChapter 13: ML Operations: The precheck script now only runs on qmd and bib filesβββββChapter 14: On-Device Learning: The precheck script now only runs on qmd and bib filesβββββChapter 18: Robust AI: The precheck script now only runs on qmd and bib filesβββββChapter 15: Security & Privacy: The precheck now only runs on .qmd and .bib filesβββββChapter 16: Responsible AI: The precheck script now only runs on qmd and bib filesβββββChapter 17: Sustainable AI: The precheck script now only runs on qmd and bib filesβββββChapter 19: AI for Good: The code now uses _quarto.yml for library management, PNG images are cached locally, and the precheck script only runs on qmd and bib filesβββββChapter 20: Conclusion: The precheck now only runs on .qmd and .bib files
π§βπ» Labs
βββββLab: Arduino Setup: The precheck script now only runs on qmd and bib filesβββββLab: Arduino Image Classification: The precheck script now only runs on qmd and bib filesβββββLab: Arduino Keyword Spotting: The precheck script now only runs on qmd and bib filesβββββLab: Arduino Motion Classification: The precheck script now only runs on qmd and bib filesβββββLab: XIAO Setup: The precheck script now only runs on qmd and bib filesβββββLab: XIAO Image Classification: The precheck script now only runs on qmd and bib filesβββββLab: XIAO Object Detection: The precheck script now only runs on qmd and bib filesβββββLab: XIAO Keyword Spotting: The precheck script now only runs on qmd and bib filesβββββLab: XIAO Motion Classification: The precheck script now only runs on .qmd and .bib filesβββββLab: Raspi: The precheck script now only runs on qmd and bib filesβββββLab: Raspberry Pi Setup: The precheck script now only runs on qmd and bib filesβββββLab: Pi Image Classification: The precheck script now only runs on qmd and bib filesβββββLab: Pi Object Detection: The precheck script now only runs on qmd and bib filesβββββLab: Pi Large Language Models: The precheck script now only runs on qmd and bib filesβββββLab: Pi Vision Language Models: The precheck script now only runs on qmd and bib filesβββββLab: Shared: The precheck script now only runs on qmd and bib filesβββββLab: Kws Feature Eng: The precheck script now only runs on qmd and bib filesβββββLab: Dsp Spectral Features Block: The precheck script now only runs on qmd and bib filesβββββLab: Lab Setup: The precheck script now only runs on .qmd and .bib filesβββββLab: Labs Overview: The precheck script is updated to run only on .qmd and .bib files
π Appendix
βββββPhD Survival Guide: The precheck now only runs on .qmd and .bib files
February 02
π Frontmatter
βββββAcknowledgements: The acknowledgements section now includes a list of contributors
π Chapters
βββββChapter 12: Benchmarking AI: The chapter now features improved learning objectives, a corrected table, updated references, reorganized content, and an updated case studyβββββChapter 1: Introduction: All ###* callout titles have been converted to a title block formatβββββChapter 2: ML Systems: All callout ###* titles have been updated to a title block formatβββββChapter 3: DL Primer: All callout ###* titles have been updated to use a title block formatβββββChapter 4: DNN Architectures: Errata were fixed and all callout ###* titles were updated to a title block formatβββββChapter 5: AI Workflow: All ###* callout titles have been updated to a title block formatβββββChapter 6: Data Engineering: All callout ###* titles have been updated to a title block formatβββββChapter 7: AI Frameworks: The way TikZ packages are used in the chapter has been restructured by moving their declaration outside of begin{..} blocks and into a header fileβββββChapter 8: AI Training: Figure 8.8 was updated and several new diagrams were added to enhance visual understanding. Python code was also corrected for accuracyβββββChapter 9: Efficient AI: Callout titles were updated to a title block format, the bibliography was updated, and an incorrect entry was removedβββββChapter 10: Model Optimizations: All ###* callout titles have been updated to use a title block formatβββββChapter 11: AI Acceleration: All ###* callout titles have been updated to use a title block formatβββββChapter 13: ML Operations: All ###* callout titles have been updated to use a title block formatβββββChapter 14: On-Device Learning: All callout ###* titles have been updated to a title block formatβββββChapter 18: Robust AI: All callout ###* titles have been updated to use a title block formatβββββChapter 15: Security & Privacy: All callout ###* titles have been updated to use a title block formatβββββChapter 16: Responsible AI: All callout ###* titles have been updated to a title block formatβββββChapter 17: Sustainable AI: All ###* callout titles have been updated to a title block formatβββββChapter 19: AI for Good: All callout ###* titles have been updated to a title block format
September 20
π Chapters
βββββChapter 11: AI Acceleration: Character formatting was fixed, figure references were corrected, and all chapters are now completeβββββChapter 5: AI Workflow: References were fixed and all chapters were revisedβββββChapter 3: DL Primer: Fixed a broken link to a video and corrected instances of β being used instead of β, while also completing revisions for all chaptersβββββChapter 6: Data Engineering: Character formatting was fixed, formatting was improved, and figure references were addedβββββChapter 1: Introduction: Figure references were fixed and all references to embedded AI were removedβββββChapter 9: Efficient AI: Figure references were fixed and all chapters are now completeβββββChapter 14: On-Device Learning: References were fixed, a path was corrected, and all chapters are now completeβββββChapter 7: AI Frameworks: Figure references were fixed and all chapters are now completeβββββChapter 13: ML Operations: Figure references were fixed and all chapters were updatedβββββChapter 15: Security & Privacy: Merge conflicts were resolved and figure references were correctedβββββChapter 12: Benchmarking AI: The chapterβs formatting was improved, some figures were removed, and merge conflicts were resolvedβββββChapter 19: AI for Good: References and figure citations were correctedβββββChapter 10: Model Optimizations: Corrected single quotes to apostrophes throughout the chapterβββββChapter 8: AI Training: Corrected instances where single quotation marks were used inconsistently with apostrophesβββββChapter 17: Sustainable AI: Figure references were corrected and typos in the sustainability section were fixedβββββChapter 2: ML Systems: Figure references were corrected throughout the chapter
π§βπ» Labs
βββββLab: Arduino Motion Classification: The lab code has been fixed by BravoBaldoβββββLab: XIAO Setup: Fixes were implemented to improve the XIAO ESP32S3 setup processβββββLab: XIAO Image Classification: The output format for the lab has been fixedβββββLab: XIAO Object Detection: The lab received fixes and improvements contributed by BravoBaldoβββββLab: XIAO Keyword Spotting: Fixes were implemented to improve the accuracy and efficiency of the keyword spotting modelβββββLab: Raspberry Pi Setup: Character formatting was fixed and additional files were added to the lab setupβββββLab: Lab Setup: Lab setup instructions have been corrected for accuracy and clarityβββββLab: Arduino Object Detection: Fixes were implemented by BravoBaldoβββββLab: XIAO Motion Classification: An image was fixedβββββLab: Raspi: Fixes were implemented by BravoBaldoβββββLab: Pi Image Classification: Character formatting was fixed and typos were correctedβββββLab: Pi Object Detection: Typos have been corrected
September 12
π Chapters
βββββChapter 13: ML Operations: The chapter content has been revised based on feedback from BaldoβββββChapter 18: Robust AI: The chapter now includes corrections and improvements addressing robustness concerns in AI modelsβββββChapter 20: Conclusion: The conclusion has been revised based on feedbackβββββChapter 19: AI for Good: The chapter contains various fixesβββββChapter 17: Sustainable AI: A quoteβs indentation was fixed and formatting was improved throughout the chapter
π§βπ» Labs
βββββLab: Pi Object Detection: The Object Detection Lab files have been uploaded for use with Raspberry PiβββββLab: Pi Image Classification: A link and some typos were corrected
September 06
π Chapters
βββββChapter 16: Responsible AI: The bibliography and text were corrected for accuracy
September 04
π Chapters
βββββChapter 11: AI Acceleration: The chapter contains grammar fixes and improved explanations for clarityβββββChapter 9: Efficient AI: The explanations in Chapter 9 have been refined for better clarityβββββChapter 1: Introduction: Captions for even side pages have been correctedβββββChapter 15: Security & Privacy: The chapter received bug fixes and grammar improvementsβββββChapter 19: AI for Good: Grammar errors were corrected throughout the chapterβββββChapter 12: Benchmarking AI: Grammar errors were corrected throughout the chapterβββββChapter 20: Conclusion: Grammar errors were corrected throughout the conclusion chapterβββββChapter 6: Data Engineering: Grammar errors were corrected throughout the chapterβββββChapter 3: DL Primer: Grammar errors were corrected for improved readabilityβββββChapter 7: AI Frameworks: The chapter contains grammar fixes for improved clarityβββββChapter 2: ML Systems: The chapter contains grammar fixes for improved clarityβββββChapter 14: On-Device Learning: Grammar errors were corrected throughout the chapterβββββChapter 13: ML Operations: Grammar errors were corrected throughout the chapterβββββChapter 10: Model Optimizations: Grammar errors were corrected throughout the chapterβββββChapter 16: Responsible AI: Grammar errors were corrected throughout the chapterβββββChapter 18: Robust AI: Grammar errors were corrected throughout the chapterβββββChapter 17: Sustainable AI: Grammar errors were corrected throughout the chapterβββββChapter 8: AI Training: Grammar errors were corrected throughout the chapter
π§βπ» Labs
βββββLab: Arduino Image Classification: The lab now contains grammar corrections for improved readabilityβββββLab: Kws Feature Eng: Grammar was corrected throughout the lab
September 02
π Chapters
βββββChapter 13: ML Operations: The ML Operations chapter now includes a new section on model serving and has been updated with improved content and corrected errorsβββββChapter 11: AI Acceleration: The chapter now includes a table fix, more student-focused explanations of hardware design principles, and an introductionβββββChapter 2: ML Systems: A dangling sentence was corrected
π§βπ» Labs
βββββLab: Raspberry Pi Setup: The Raspberry Pi setup instructions were updated and additional files were included for the lab setupβββββLab: Pi Image Classification: Files were added to the lab
August 29
π Chapters
βββββChapter 13: ML Operations: The ML Operations chapter has been updated with corrections based on feedbackβββββChapter 14: On-Device Learning: The chapter received fixes based on feedback from (BravoBaldo?)
π§βπ» Labs
βββββLab: Labs: Resolved an issue with merging tablesβββββLab: Pi Image Classification: Files were added to the labβββββLab: Kws Feature Eng: Formatting issues were fixed
August 27
π Chapters
βββββChapter 10: Model Optimizations: The knowledge distillation explanation was improved, the challenges were adjusted to be less repetitive and more informative, and the explanations of structure importance methods were correctedβββββChapter 9: Efficient AI: Duplicate information between chapters 8 and 9 was removed, background about floating point representation was added, and explanations of structure importance methods were correctedβββββChapter 12: Benchmarking AI: Benchmarking AI chapter now includes corrections for improved accuracy and clarityβββββChapter 7: AI Frameworks: Fixed broken links and improved table formatting in the AI Frameworks chapterβββββChapter 15: Security & Privacy: The Power Attack and Side-Channel Attack sections were edited and broken links were fixedβββββChapter 11: AI Acceleration: This update addresses several issues including broken links, inaccurate information about qbits, hyphen errors, a duplicate title, and a grammatically incorrect sentenceβββββChapter 13: ML Operations: Minor wording changes were made for improved clarity and flowβββββChapter 17: Sustainable AI: Broken links in the chapter have been fixed
π§βπ» Labs
βββββLab: Xiao Esp32S3: Table formatting for grid tables has been fixed
August 22
π Chapters
βββββChapter 19: AI for Good: A section on AI for climate change mitigation was added, the disaster response case study was updated, and new ethical considerations related to AI for good were includedβββββChapter 11: AI Acceleration: Added a section on systolic arrays for matrix multiplication, updated the discussion of tensor cores to reflect the latest NVIDIA architectures, and clarified the impact of memory hierarchy on AI performanceβββββChapter 17: Sustainable AI: A new section on ethical considerations in sustainable AI was added, existing content on data efficiency techniques was updated, and a case study on responsible AI development was included
π§βπ» Labs
βββββLab: Labs: Lab files were added to the textbookβββββLab: Raspi: Files were added to the Raspi labβββββLab: Raspberry Pi Setup: Files were added to the Raspberry Pi Setup lab
August 21
π Chapters
βββββDsp Spectral Features Block: Removed unused code from the chapterβββββLab: Arduino Image Classification: Dead code was removed from the labβββββChapter 10: Model Optimizations: Typos were corrected throughout the chapterβββββChapter 6: Data Engineering: Typos were fixedβββββChapter 7: AI Frameworks: Typos were fixed throughout the chapterβββββChapter 11: AI Acceleration: Typos were fixedβββββChapter 14: On-Device Learning: Typos were correctedβββββChapter 13: ML Operations: Typos were correctedβββββChapter 15: Security & Privacy: Typos were fixedβββββChapter 8: AI Training: Inconsistent bold text was removed for improved readabilityβββββChapter 2: ML Systems: A minor typo was corrected
π§βπ» Labs
βββββLab: Raspi: This lab introduces a hands-on experience with Raspberry Pi for machine learning applicationsβββββLab: Raspberry Pi Setup: This lab now includes initial instructions for setting up a Raspberry Pi for machine learning tasksβββββLab: Pi Image Classification: This lab introduces a basic image classification pipeline using a Raspberry PiβββββLab: Pi Object Detection: This lab introduces object detection using a Raspberry PiβββββLab: Labs: The lab section now includes a fixed table format and an initial version of the rasPi labβββββLab: Pi Large Language Models: This lab introduces a basic implementation of a large language model on a Raspberry PiβββββLab: Lab Setup: The lab now includes initial instructions for setting up a Raspberry Pi environment
August 15
π Chapters
βββββChapter 11: AI Acceleration: Table references were fixed for accuracyβββββChapter 13: ML Operations: Table references were fixedβββββChapter 15: Security & Privacy: Table references were fixed for accuracy
August 15
π Chapters
βββββChapter 6: Data Engineering: Several pivotal fixes and enhancements were made to clarify and improve the data engineering contentβββββChapter 18: Robust AI: A citation reference was fixed and tables were given striped and hover stylesβββββChapter 13: ML Operations: Tables now have striped and hover styling, centering has been fixed, and the grid table has been updatedβββββChapter 15: Security & Privacy: The table in this chapter has been updated to a grid format and several fixes have been implemented for clarity and accuracyβββββChapter 8: AI Training: A reference was corrected, a broken Colab link was fixed, and table formatting was updatedβββββChapter 19: AI for Good: The AI for Good chapter has been enhancedβββββChapter 12: Benchmarking AI: The chapter now includes fixes for existing code examples and enhanced explanations of various benchmarking metricsβββββChapter 3: DL Primer: The chapter includes various enhancements and bug fixes for improved clarity and accuracyβββββChapter 9: Efficient AI: Several errors were fixed and explanations were improved for clarityβββββChapter 7: AI Frameworks: The chapter now includes fixes related to delve and enhancements that utilize framework examplesβββββChapter 11: AI Acceleration: The chapterβs tables were converted into grid tables and updated for clarityβββββLab: Arduino Image Classification: Improved Arduino code examples and corrected small errors in the image classification implementationβββββChapter 1: Introduction: Several issues were fixed and explanations were enhanced for clarityβββββChapter 2: ML Systems: Implemented pivotal fixes and enhancements to improve clarity and accuracyβββββChapter 14: On-Device Learning: The chapter now includes a grid table and has been corrected for errorsβββββChapter 10: Model Optimizations: The chapterβs content has been refined with corrections, clearer explanations, and updated terminologyβββββChapter 16: Responsible AI: The table in this chapter has been updated to a grid formatβββββChapter 17: Sustainable AI: The chapter now includes various fixes, improved utilization of concepts, and a more in-depth exploration of existing fixesβββββChapter 5: AI Workflow: The AI Workflow chapter now includes several fixes and expanded explanations of key concepts
π§βπ» Labs
βββββLab: Arduino Image Classification: The Arduino Image Classification lab has been enhancedβββββLab: XIAO Image Classification: Delve fixes were implementedβββββLab: XIAO Keyword Spotting: The lab now utilizes
August 15
π Chapters
βββββChapter 8: AI Training: The chapter now includes a discussion of regularization techniques and hyperparameter search methods, consolidates information on common training pitfalls, and updates the neural network notationβββββChapter 9: Efficient AI: The chapter was updated with changes incorporated from the βdevβ branchβββββChapter 7: AI Frameworks: Typos were corrected throughout the chapter
August 06
π Chapters
βββββChapter 7: AI Frameworks: The chapter now focuses more on advanced features while removing the federated learning section. Table formatting has been improved for better readability and consistencyβββββDsp Spectral Features Block: Resolved image width problems to ensure proper display in PDF formatβββββLab: Arduino Image Classification: Image width issues were fixed for PDF rendering, redundant underlines were removed, and all broken video links were repairedβββββChapter 13: ML Operations: The chapter now features corrected URL links, updated table formatting for a grid layout, and consistent credit attribution styleβββββChapter 18: Robust AI: Tables were updated with grid formatting, source attributions were revised, and builds for HTML and PDF were implementedβββββChapter 5: AI Workflow: Tables now use grid formatting and are left-aligned, incorporating student feedbackβββββChapter 3: DL Primer: Fixed all broken URL links in the chapter, updated the βCreditβ section to be consistent with βSourceβ, and added grid tables formatted with left alignmentβββββChapter 6: Data Engineering: The text was updated slightly and a new Wake Vision Colab exercise was addedβββββChapter 15: Security & Privacy: The privacy and security section was edited and broken links were fixedβββββChapter 19: AI for Good: Fixed all broken URL links, updated credit sources with consistent formatting, and added HTML and PDF builds compatible with Edward Tufteβs style guideβββββChapter 11: AI Acceleration: All broken URL links were fixed and the credit source formatting was updated to be consistentβββββChapter 14: On-Device Learning: All broken URL links were fixed and table formatting was updated to use left alignment with markdown. The βCreditβ section was renamed to βSourceβ and the formatting style was made consistentβββββChapter 16: Responsible AI: All broken URL links were fixed, credit sources were updated with consistent formatting, and an HTML + PDF build was added for Edward TufteβββββChapter 8: AI Training: All broken URL links were fixed, source citation formatting was updated, and grid tables with left-aligned markdown formatting were addedβββββChapter 9: Efficient AI: Grid tables were added to the chapter, all tables are now left-aligned, and image paths and figure IDs were updated for consistencyβββββChapter 2: ML Systems: Grid tables were added to the chapter, all tables are now left-aligned, and source attributions were updated with consistent formattingβββββChapter 10: Model Optimizations: Implemented grid tables for better visualization, aligned all tables with markdown formatting, and added in-text citations for sourcesβββββChapter 17: Sustainable AI: Source citations were updated and formatting was made consistent across the chapter. Additionally, HTML and PDF builds were createdβββββChapter 1: Introduction: The textbook now has HTML and PDF output formatsβββββChapter 12: Benchmarking AI: Credit attributions were updated and formatting was corrected for consistency
π§βπ» Labs
βββββLab: Nicla Vision: Tables are now left-aligned and credit information is consistently formatted as βSourceββββββLab: Arduino Setup: Image width issues were fixed to improve PDF rendering and redundant underlines were removedβββββLab: Arduino Image Classification: Image width issues were fixed for PDF rendering, redundant underlines were removed, and remaining broken video links were correctedβββββLab: Arduino Object Detection: Image width issues for PDF rendering and broken video links have been fixedβββββLab: Arduino Keyword Spotting: Image width issues were fixed to improve PDF rendering and all broken URL links were correctedβββββLab: Arduino Motion Classification: Image width issues affecting PDF rendering have been fixedβββββLab: XIAO Setup: Image widths were adjusted to prevent rendering issues in PDFsβββββLab: XIAO Image Classification: Image width issues affecting PDF rendering have been fixedβββββLab: XIAO Object Detection: Image width issues affecting PDF rendering have been fixedβββββLab: XIAO Keyword Spotting: Image width issues were fixed for better PDF rendering and broken links were correctedβββββLab: XIAO Motion Classification: Image width issues that affected PDF rendering have been fixedβββββLab: Dsp Spectral Features Block: Image width issues affecting PDF rendering have been fixedβββββLab: Xiao Esp32S3: Credit attributions were updated and formatting was made consistentβββββLab: Shared: All tables in this lab now use markdown formatting with left alignmentβββββLab: Kws Feature Eng: Image widths were adjusted to ensure proper rendering in PDFs
June 25
π Chapters
βββββChapter 3: DL Primer: The link for video 3.1 has been corrected
June 20
π Chapters
βββββChapter 2: ML Systems: The chapter now includes corrections to a broken reference and incorporates feedback from students on the βdevβ branch
π§βπ» Labs
βββββLab: Shared: Broken links were fixed
June 19
π Chapters
βββββObject Detection Fomo: Labs have been integrated into the Object Detection Fomo chapterβββββChapter 12: Benchmarking AI: A figure illustrating MLPerf training progress was added, and the chapter content was updated based on feedback. Formatting and line spacing were also correctedβββββChapter 11: AI Acceleration: Citation formatting was changed from parentheses to brackets, MD lint fixes were implemented, and a link to Googleβs Edge TPU website was addedβββββChapter 1: Introduction: The introduction material was updated and the foreword content was removed to correct an errorβββββChapter 2: ML Systems: The chapter content has been refined and improved based on feedback and data reviewβββββChapter 6: Data Engineering: Citation formatting was changed from parentheses to square brackets and minor style fixes were madeβββββChapter 10: Model Optimizations: Citation formatting was changed from parentheses to brackets and several typos were fixedβββββChapter 18: Robust AI: Citation formatting was changed from parentheses to brackets and markdown linting issues were addressedβββββAcknowledgements: The acknowledgements were updated with MD lint fixes and some comments were disabled on specific pagesβββββChapter 19: AI for Good: Applied Markdown linting fixes for improved formatting and consistencyβββββChapter 20: Conclusion: Applied Markdown linting fixes for consistency and readabilityβββββChapter 3: DL Primer: The chapter contains a batch of typographical errors and formatting issues that have been correctedβββββChapter 9: Efficient AI: Markdown lint errors were fixed and a missing video link was addedβββββChapter 7: AI Frameworks: The chapter contains a batch of typo and format fixesβββββLab: Arduino Image Classification: The lab notebook underwent code style and formatting improvements using MD lintβββββChapter 14: On-Device Learning: Applied markdown linting fixes to the chapterβββββChapter 13: ML Operations: Applied Markdown linting fixes for consistency and styleβββββChapter 15: Security & Privacy: The case study header was fixed and some markdown linting issues were resolvedβββββChapter 16: Responsible AI: The Markdown code was cleaned up using a linting toolβββββChapter 17: Sustainable AI: Applied Markdown linting fixes for improved formatting and consistencyβββββChapter 8: AI Training: This chapter contains a batch of typo and format fixesβββββChapter 5: AI Workflow: The Markdown files were cleaned up using a linting toolβββββDsp Spectral Features Block: The resources were fixed and the title was adjustedβββββGenerative Ai: The wording of several passages has been refined for clarity
π§βπ» Labs
βββββLab: Labs: The lab section was reorganized, the overview was updated, a table was transposed and updated, and labs were integrated into the main textβββββLab: Arduino Image Classification: This update integrates all Arduino lab files into the curriculum and fixes resource linksβββββLab: Nicla Vision: The lab now includes KWS, has improved introductory text, updated images, and corrected build errorsβββββLab: Arduino Setup: Resources were fixed, some elements were renamed, and the setup content was integrated into the lab structureβββββLab: Arduino Object Detection: Resource links were fixed and the object detection lab was integrated into the existing labs sectionβββββLab: Arduino Keyword Spotting: The lab now has correctly placed content, improved grammar (βcodesβ changed to βcodeβ), and updated resourcesβββββLab: Arduino Motion Classification: Content placement was corrected, grammar was improved, and the lab was integrated into the curriculumβββββLab: Xiao Esp32S3: The lab now has a streamlined introduction, corrected build errors, includes proper image attribution, and is better integrated within the overall lab structureβββββLab: XIAO Setup: Resources were fixed, SEEED labs were imported, and files were renamedβββββLab: XIAO Image Classification: The lab now imports SEEED labs and integrates them into the existing content, with resource links also being fixedβββββLab: XIAO Object Detection: The lab now incorporates resources and imports from the SEEED labs while addressing MD lint issuesβββββLab: XIAO Keyword Spotting: Resources were fixed, SEEED labs were imported and integrated into the lab exercisesβββββLab: XIAO Motion Classification: This lab now includes links to internal documentation and integrates with other SEEED labsβββββLab: Shared: Added an overview section for the shared labsβββββLab: Kws Feature Eng: The lab now includes MD lint fixes, updated resources, and is integrated into the broader lab structureβββββLab: Dsp Spectral Features Block: The lab blockβs title was slightly adjusted and integrated into the broader lab environmentβββββLab: Lab Setup: Added getting started content to the lab setup and updated the overview section with a placeholder for detailed information
June 11
π Chapters
βββββChapter 18: Robust AI: The chapter now includes video callouts, reorganized exercise callouts, updated references and links, an expanded general description, and information about Bayesian Neural NetworksβββββChapter 11: AI Acceleration: Video callouts were added to the chapter along with end-of-section resource recommendations, and formatting and spacing issues were resolvedβββββChapter 19: AI for Good: Video callouts were added to the chapter and exercise callout blocks were folded for improved visual appealβββββChapter 12: Benchmarking AI: Video callouts were added to the chapter, a previous .callout issue was fixed, and spacing and formatting of exercise callouts were improvedβββββChapter 6: Data Engineering: Video callouts were added to the chapter and exercise callout blocks were folded for improved visual appealβββββChapter 3: DL Primer: Video callouts were added and end-of-section resources were updatedβββββChapter 9: Efficient AI: Video callouts and end-of-section resources were added, and a previous .callout issue was fixedβββββChapter 7: AI Frameworks: Video callouts were added to the chapter and exercise callout blocks were folded for improved aestheticsβββββChapter 2: ML Systems: Video callouts were added to the chapter and exercise callout blocks were folded for improved visual presentationβββββChapter 14: On-Device Learning: Video callouts were added and exercise callout blocks were folded for improved visual presentationβββββChapter 13: ML Operations: Video callouts were added to the chapter and exercise callout blocks were folded for improved visual appealβββββChapter 10: Model Optimizations: Video callouts were added to the chapter and exercise callout blocks were folded for improved visual presentationβββββChapter 15: Security & Privacy: Video callouts were added and exercise callout blocks were folded for improved visual presentationβββββChapter 16: Responsible AI: Video callouts and end-of-section resources were added to the chapterβββββChapter 17: Sustainable AI: Video callouts and exercise callout blocks were added and formatted for improved readabilityβββββChapter 8: AI Training: Video callouts were added, exercise callout blocks were folded for improved aesthetics, and a table rendering issue was fixedβββββChapter 5: AI Workflow: Video callouts were added to the chapter and a previous formatting issue with .callout was fixedβββββGenerative Ai: Added text about future generative AI content and made minor edits to existing βcoming soonβ sections
June 01
π Chapters
βββββChapter 19: AI for Good: The Colab badge was fixed, the βcoming soonβ section now uses bullets, and the slide notes use the default format for better PDF renderingβββββChapter 12: Benchmarking AI: The Colab badge was fixed, the βcoming soonβ section now uses bullets, and the default note for slides was implemented for better PDF renderingβββββChapter 6: Data Engineering: The Colab badge was fixed, the βcoming soonβ section was updated with bullets, and minor markdown issues were corrected in the text and URL highlightingβββββChapter 3: DL Primer: The Colab badge is now fixed, the βcoming soonβ section uses bullets instead of a paragraph, and the default note for slides has been changedβββββChapter 7: AI Frameworks: The Colab badge was fixed, the βcoming soonβ section was updated with bullets, and the default slide note was changed to render better in PDFsβββββChapter 11: AI Acceleration: The Colab badge was fixed, the βcoming soonβ section was updated with bullets, and the default note for slides was changed to render better in PDFβββββChapter 2: ML Systems: The Colab badge was fixed, rendering issues were resolved, and the βcoming soonβ sectionβs formatting was updatedβββββChapter 14: On-Device Learning: The Colab badge was fixed, the βcoming soonβ section was updated to use bullets, and slide note formatting was changed for better PDF renderingβββββChapter 13: ML Operations: The Colab badge was fixed, the βcoming soonβ section now uses bullets, and the default note for slides was changed for better PDF renderingβββββChapter 10: Model Optimizations: The Colab badge was repaired, the βcoming soonβ section was updated to use bullets, and the default note for slides was changed to render better in PDFβββββChapter 15: Security & Privacy: The Colab badge was fixed, the βcoming soonβ section now uses bullets, and the slide notes use the default format for better PDF renderingβββββChapter 18: Robust AI: Resolved Colab badge display problems, implemented (exr?)- notation for Colab notebooks, and optimized slide rendering for PDFsβββββChapter 17: Sustainable AI: The Colab badge was fixed, the βcoming soonβ section was updated with bullets, and formatting was improved for labs, exercises, and slidesβββββChapter 8: AI Training: The Colab badge was fixed, the βcoming soonβ section was updated to use bullets, and the default slide note was changed to render better in PDFsβββββChapter 9: Efficient AI: Fixed rendering problems, corrected some headers, and set a default note for slides to ensure proper PDF renderingβββββChapter 16: Responsible AI: The coming soon section now uses bullets, slide notes utilize the default format for better PDF rendering, and formatting was adjusted in labs, exercises, and slidesβββββChapter 5: AI Workflow: The coming soon section now uses bullets, the slide default note was changed for better PDF rendering, and formatting was adjusted in labs, exercises, and slidesβββββChapter 1: Introduction: Reference links were re-added to the introduction section and the grammar and readability were improvedβββββGenerative Ai: Fixed rendering problems in the chapter
May 26
π Chapters
βββββChapter 10: Model Optimizations: This update includes added videos, updated exercises, and additional slides for a more comprehensive learning experienceβββββChapter 18: Robust AI: The Robust AI chapter now includes learning objectives and updated figures, incorporating feedback from Yanjing for improvementsβββββChapter 14: On-Device Learning: This chapter now includes videos, additional collaborative exercises, and updated content based on recent slidesβββββChapter 20: Conclusion: The conclusion section was updated and a cover image was added to the fileβββββChapter 3: DL Primer: This chapter now includes short captions for all videos, an updated Data Diversity and Quality section, and a fix for SVG to PNG conversion to enable PDF buildsβββββChapter 17: Sustainable AI: The conclusion was updated, figure captions and references were fixed, and section headers were added for cross-referencingβββββChapter 9: Efficient AI: Section headers were added for easier cross-referencing, additional slides were incorporated, and figure captions and references were correctedβββββChapter 6: Data Engineering: The chapter now includes additional slides and updated content with exercises. The colab-badge.svg was also converted from SVG to PNG to enable PDF buildsβββββChapter 8: AI Training: Video captions were added, section headers were implemented for easier cross-referencing, and figure captions and references were correctedβββββChapter 7: AI Frameworks: Fixed broken slide links, added section headers for easier cross-referencing, and updated the frameworks.qmd fileβββββChapter 13: ML Operations: Short captions were added for videos, the ops.qmd file was updated, and the colab-badge.svg was changed from SVG to PNGβββββChapter 15: Security & Privacy: All tables now have captions, relative path links were cleaned up, and short captions were added to the videosβββββChapter 5: AI Workflow: Slide links were fixed, section headers were added for cross-referencing, and the bib file header was correctedβββββChapter 19: AI for Good: Section headers were added for easier cross-referencing, video captions were included, and the colab-badge.svg was changed to PNG to enable PDF buildsβββββChapter 12: Benchmarking AI: Section headers were added for easier cross-referencing, punctuation was corrected throughout the chapter, and the colab-badge.svg was changed to PNG to enable PDF buildsβββββChapter 11: AI Acceleration: Added section headers for easier cross-referencing, fixed broken links, and updated figure captions and referencesβββββChapter 2: ML Systems: Added section headers for cross-referencing, updated names throughout the chapter, and removed the βEmbedded Systemsβ section to focus solely on ML systemsβββββChapter 16: Responsible AI: Section headers were added for easier cross-referencing, a broken link was fixed, and grammatical errors were correctedβββββGenerative Ai: Added section headers for easier cross-referencing and updated βComing soonβ text to βComing soonββββββChapter 1: Introduction: Section headers were added for easier cross-referencing, a bibliography file header error was corrected, and a cover image was added to the introduction chapterβββββEmbedded Ml: Chapters 1 through 4 were edited and the collapse setting for the Embedded ML chapter was set to falseβββββEmbedded Sys: Chapters 1 through 4 were edited, additional slides were included, and the collapse setting was changedβββββAcknowledgements: Image logos were added and an initial draft of the acknowledgements chapter was createdβββββDsp Spectral Features Block: Corrected punctuation and replaced instances of 's with βsβββββLab: Arduino Image Classification: Punctuation has been corrected in the labβββββKws Feature Eng: Punctuation has been corrected throughout the chapterβββββMotion Classify Ad: Fixed punctuation throughout the chapterβββββNiclav Sys: Corrected punctuation throughout the chapter
March 21
π Chapters
βββββChapter 6: Data Engineering: The Resources section now includes introductory text for each part and can be collapsed, exercise callouts were added, and slides have been moved to the end of the pageβββββChapter 3: DL Primer: The Resources section now includes introductory text for each part and a collapsible feature, additional slides were incorporated, and the placement of slides was moved to the end of the pageβββββChapter 19: AI for Good: A βResourcesβ section was added to the end of all QMD files, and the existing Resources were updated with introductory text and enabled to be collapsedβββββChapter 12: Benchmarking AI: A βResourcesβ section was added to the end of all QMD files with collapsible sections for each part and the slides were moved to the end of the pageβββββChapter 9: Efficient AI: The Resources section in Chapter 9 now includes introductory text for each part and can be collapsed. Additional slides were added, and an empty βResourcesβ section with headers was added to the end of all QMD filesβββββEmbedded Ml: The Embedded Ml chapter now includes additional slides placed at the end of each page, and a new Resources section with introductory text for each partβββββEmbedded Sys: The Resources section now includes introductory text for each part and can be collapsed, additional slides were added, and slides have been moved to the end of the pageβββββChapter 7: AI Frameworks: The Resources section now includes introductory text for each part and collapsible elements, while slides have been moved to the end of the page. Colab badges have also been addedβββββChapter 11: AI Acceleration: The Resources section now includes introductory text for each part and can be collapsed. Empty βResourcesβ sections were added to the end of all QMD files with headersβββββChapter 14: On-Device Learning: The Resources section now includes introductory text for each part and can be collapsed, additional slides were added, and all slides have been moved to the end of the pageβββββChapter 13: ML Operations: Resource sections now have introductory text and collapsible functionality, additional slides were incorporated, and slides were relocated to the end of pagesβββββChapter 10: Model Optimizations: The Resources section now includes introductory text for each part and has been moved to the end of pages. Empty βResourcesβ sections have also been added to all QMD filesβββββChapter 15: Security & Privacy: The Resources section now includes introductory text for each part and can be collapsed, additional slides were added, and slides have been moved to the end of the pageβββββChapter 16: Responsible AI: A βResourcesβ section was added to the end of the chapter with collapsible functionality and slides were moved to the end of the pageβββββChapter 17: Sustainable AI: The Resources section now includes introductory text for each part and slides have been moved to the end of the page. An empty βResourcesβ section has also been added to all QMD filesβββββChapter 8: AI Training: Resource sections now include introductory text and can be collapsed, additional slides were incorporated, and all slides are now located at the end of the pageβββββChapter 5: AI Workflow: Each resource section now has introductory text, slides have been moved to the end of pages, and a new βResourcesβ section has been added to all QMD files
March 12
π Chapters
βββββChapter 7: AI Frameworks: Added Colab notebooks to the chapter and included additional slides while also fixing previous notesβββββChapter 6: Data Engineering: The data engineering chapter now includes Colab notebooks, a new web scraping exercise in both the subsection and as a separate part of the exercises, and additional slidesβββββChapter 11: AI Acceleration: Removed a figure reference and a Mermaid diagram section, and made fixes to the non-ASCII checker scriptsβββββNiclav Sys: Corrected broken links and implemented non-ASCII checking scripts with fixes for related issuesβββββEmbedded Ml: Removed debug code, added a nested example, and implemented arrow capability and custom callouts in the style fileβββββChapter 19: AI for Good: The chapter now includes additional slides and incorporates non-( ) ASCII checker scripts with accompanying fixesβββββChapter 12: Benchmarking AI: The chapter now includes additional slides, corrected notes from the previous week, and incorporates non-ASCII checker scriptsβββββChapter 3: DL Primer: The chapter now includes additional slides and incorporates non-ASCII checking scripts for improved accuracyβββββEmbedded Sys: The chapter now includes additional slides and incorporates non-ASCII checking scripts with accompanying fixesβββββChapter 14: On-Device Learning: The chapter content was expanded with additional slides and included non-( ) ASCII checker scripts along with fixesβββββChapter 13: ML Operations: More slides were added to the chapter, last weekβs notes were fixed, and scripts for checking non-( ) ASCII were addedβββββChapter 15: Security & Privacy: The chapter content was expanded with additional slides and corrections were made to previous notesβββββChapter 16: Responsible AI: The Responsible AI chapter now includes additional slides and has had its style corrected to ensure proper renderingβββββChapter 17: Sustainable AI: Added more slides and implemented non-ASCII checking scripts with accompanying fixesβββββChapter 8: AI Training: The chapter now includes additional slides, non-( ) ASCII checker scripts have been added and fixed, and style issues affecting successful rendering have been resolvedβββββChapter 5: AI Workflow: The chapter includes additional slides, corrected previous notes, and incorporates non-(ASCII) checker scripts for improved accuracyβββββChapter 10: Model Optimizations: Non-ASCII checker scripts were added and style issues were fixed to ensure successful renderingβββββChapter 9: Efficient AI: Incorporated non-ASCII checker scripts and resolved problems with non-ASCII characters in code examples
February 03
π Chapters
βββββChapter 12: Benchmarking AI: Unpopulated list items were removed and the styling of list items was made more consistentβββββChapter 13: ML Operations: The MCU example for a smartwatch was updated and a new reference was includedβββββChapter 15: Security & Privacy: Video rendering is fixed, GDPR and CCPA section hyperlinking is improved, and the CCPA summary is clearerβββββChapter 11: AI Acceleration: Video rendering issues in the chapter have been resolvedβββββChapter 3: DL Primer: Video rendering issues in the chapter have been fixedβββββChapter 19: AI for Good: Video rendering and YouTube shortened URL resolution issues were fixedβββββChapter 14: On-Device Learning: The rendering of itemized lists has been fixedβββββChapter 17: Sustainable AI: List item spacing was made consistent throughout the chapter and a reference to the OECD blueprint paper was added
February 02
π Chapters
βββββChapter 19: AI for Good: Several image references were fixed, all bibtex entries were updated automatically, and the β81-figure-references/part-2β branch was mergedβββββChapter 11: AI Acceleration: Several image links were repaired, bibliographic information was corrected, and all BibTeX references were updated automaticallyβββββChapter 13: ML Operations: Several image references were fixed, all bibtex entries were updated automatically, and figure 14.3 now renders correctlyβββββChapter 15: Security & Privacy: Several image references were fixed, grammar was corrected, and video URLs were adjusted to ensure proper renderingβββββChapter 17: Sustainable AI: Several image links were fixed, all BibTeX references were updated automatically, and a broken chapter link was correctedβββββChapter 6: Data Engineering: Modified callout content, added a web scraping Colab exercise, updated all bibtex references automatically, and converted PNG images to JPG formatβββββChapter 16: Responsible AI: Redundant citation issues were fixed and all BibTeX references were updated automaticallyβββββChapter 12: Benchmarking AI: Reference rendering in the chapter is fixed and all BibTeX references have been updated automaticallyβββββChapter 10: Model Optimizations: A missing reference for quantization-aware pruning was added, an incomplete sparsity matrix filter illustration was removed and moved to a later section, and all bibtex references were updated automaticallyβββββChapter 14: On-Device Learning: The chapterβs bullet list and formatting were updated, and a broken image reference was removedβββββChapter 3: DL Primer: SVG images were removed from the chapter and replaced with PNGs to ensure proper rendering in PDF outputsβββββEmbedded Sys: Bibtex references throughout the chapter were automatically updatedβββββChapter 8: AI Training: All BibTeX references have been updated automaticallyβββββEmbedded Ml: The image format in the chapter has been switched to PNG
January 28
π Frontmatter
βββββAcknowledgements: The contributor list has been updated to include all contributors
January 28
π Frontmatter
βββββAcknowledgements: The acknowledgements section now includes logos and an updated list of contributors
π Chapters
βββββChapter 5: AI Workflow: This chapter now includes a section on data preprocessing techniques, an updated discussion on feature engineering, and a new case study on applying AI workflow to natural language processingβββββChapter 6: Data Engineering: The data section now includes more citations and edits were made to later data engineering sections. Work is ongoing on a keyword sectionβββββChapter 8: AI Training: Added figures illustrating chip concepts, updated references to align with the hardware section, and made improvements to the training chapter contentβββββChapter 19: AI for Good: The chapter includes new videos, images, updated learning objectives, and added spotlight use casesβββββChapter 2: ML Systems: Radar plots were added to Chapter 2βββββChapter 7: AI Frameworks: Figures of chips were added and small bibliographic references were includedβββββChapter 9: Efficient AI: Learning objectives were added, references were updated, and new figures were includedβββββChapter 11: AI Acceleration: References were removed from the chapterβββββChapter 1: Introduction: A redundant case study was removedβββββChapter 4: DNN Architectures: The chapterβs wording has been refined and corrected based on recommendationsβββββChapter 10: Model Optimizations: Removed dead references throughout the chapter
π§βπ» Labs
βββββLab: Pi Image Classification: Updated the code for image classification, included explanations of the model training process, and improved the clarity of the instructionsβββββLab: Pi Object Detection: Updated the object detection code with explanations for TensorFlow Lite deployment on Raspberry Pi and improved code comments
π Appendix
βββββPhD Survival Guide: A few favorite resources were added and one link was updated
January 17
π Frontmatter
βββββSocratiQ: Updated SocratiQ interface descriptions, added examples of its use for concept mapping, and improved explanations of its learning algorithmβββββAbout: The acknowledgements section was updated to include information about open-source contributionsβββββAcknowledgements: The acknowledgements section now includes a list of contributors
π Chapters
βββββChapter 3: DL Primer: A new section on generative adversarial networks (GANs) was added, along with updated examples and improved explanations of their training processβββββChapter 7: AI Frameworks: This chapter now includes an overview of AI frameworks, updated learning objectives, and a section on the history of these frameworksβββββChapter 1: Introduction: The introduction has been revised to reflect Bravoβs feedbackβββββChapter 4: DNN Architectures: A footnote was removed and a clarification about parameter storage bounds for RNNs was added. The table format was also updated to reStructuredText with footnotesβββββChapter 6: Data Engineering: Incorporated feedback provided by BravoβββββChapter 2: ML Systems: The merge resolved issues and improved PDF renderingβββββChapter 12: Benchmarking AI: A reference issue was fixed
π§βπ» Labs
βββββLab: Pi Vision Language Models: The code for VLM training was updated and explanations of hyperparameter tuning were addedβββββLab: Pi Large Language Models: Copyediting errors were corrected
January 12
π Frontmatter
βββββAcknowledgements: The contributor list was updated and PDF rendering was improved
π Chapters
βββββChapter 1: Introduction: The introduction chapter now correctly renders triple quotes in PDFs and has adjusted section headers to improve readabilityβββββChapter 2: ML Systems: A definition for hybrid ML was added and PDF rendering was improvedβββββChapter 5: AI Workflow: Zishen made fixes to improve PDF renderingβββββChapter 6: Data Engineering: Updated figure sizes and revised the data labeling section with additions from Zishen and other contributors
January 11
π Frontmatter
βββββAcknowledgements: The contributor list has been updated to include all contributors and correct any omissions
π Chapters
βββββChapter 2: ML Systems: A decision playbook framework was added and definitions were updated for each section of the chapterβββββChapter 5: AI Workflow: A grammar fix request was removedβββββChapter 6: Data Engineering: Incorporated Bravoβs suggested corrections and improvements
January 11
π Frontmatter
βββββAbout: The bookβs target audience and acknowledgements were updatedβββββAcknowledgements: The contributor list has been updated to include all contributors and correct any omissionsβββββSocratiQ: The SocratiQ file has been edited
π Chapters
βββββChapter 6: Data Engineering: The chapter now includes additional references, updated information on synthetic data and crowdsourcing, revised web scraping content, a refreshed problem definition, and an updated overviewβββββChapter 1: Introduction: Footnotes have been added to provide additional context and information within the introduction
January 09
π Frontmatter
βββββAcknowledgements: The contributor list has been updated to include all contributors and correct any omissions
π Chapters
βββββChapter 1: Introduction: The introduction has been revised based on Marcoβs feedbackβββββChapter 5: AI Workflow: Incorporated feedback from Bravo to refine the explanations of the AI workflowβββββChapter 8: AI Training: Feedback clarity and Bravoβs output format were improvedβββββChapter 11: AI Acceleration: The chapter now includes incorporated feedback from Bravo and has improved explanationsβββββChapter 16: Responsible AI: Incorporated feedback from Bravo and improved explanations within the chapterβββββChapter 6: Data Engineering: Grammar pass fix requests were removedβββββChapter 7: AI Frameworks: Grammar pass fix requests were removed
January 07
π Frontmatter
βββββForeword: The wording in the foreword has been slightly refinedβββββAcknowledgements: The acknowledgements section now includes a list of contributors
π Chapters
βββββChapter 3: DL Primer: The chapter now includes an updated figure caption, additional images and code to clarify the training loop and inference process, and a corrected typo in an equationβββββChapter 1: Introduction: The distinction between artificial intelligence (AI) and machine learning (ML) is now more clearly definedβββββChapter 4: DNN Architectures: The chapter now includes new visualization figures and updated text based on Zishenβs feedback
January 03
π Frontmatter
βββββAcknowledgements: The contributor list has been updated to include all contributors and correct any omissionsβββββSocratiQ: Corrected several typos and inconsistencies within the SocratiQ chapter text
π Chapters
βββββChapter 1: Introduction: Definitions of supervised and unsupervised learning were clarified and examples were updated for improved comprehensionβββββChapter 2: ML Systems: The explanation of data parallelism was corrected and the discussion on model serving was updated to reflect recent advancementsβββββChapter 4: DNN Architectures: Fixed various issues within the chapterβββββChapter 6: Data Engineering: Fixed a bug in the data cleaning section and updated Apache Spark code examplesβββββChapter 20: Conclusion: Fixes were applied to the conclusion chapter
January 02
π Frontmatter
βββββAcknowledgements: The acknowledgements section now includes a complete and updated list of contributors
π Chapters
βββββChapter 4: DNN Architectures: Removed unnecessary commented text and implemented suggested editsβββββChapter 20: Conclusion: Incorporated Bravoβs suggested revisions throughout the conclusionβββββChapter: Generative Ai: Dead commented-out text has been removed from the chapter
January 01
π Frontmatter
βββββForeword: The foreword now has a smoother flow, corrected sidebar formatting issues, and includes updated text contentβββββAbout: The chapter order was updated and incorporated feedback from BravoβββββSocratiQ: The chapterβs web formatting was fixed and its content was reorganizedβββββIndex: The abstract was updatedβββββAcknowledgements: The acknowledgements section now includes contributors and the readme has been updated to reflect this
π Chapters
βββββChapter: Dl Architectures: The chapter was renamed and rewritten with an emphasis on ML systems, incorporating Colab notebooks for interactive learningβββββChapter 2: ML Systems: The chapter now includes a Mobile ML section, improved connections with text in figures, and an added example system to ground conceptsβββββChapter 4: DNN Architectures: The chapter now includes a section on transformer architectures, the RNN conclusion has been updated, and tables have been corrected for formatting and referencingβββββChapter 18: Robust AI: The chapter now includes clearer explanations of gradient norms and Bayesian Neural Networks, reframes the introduction to faults, and focuses on ML-specific fault examples in edge casesβββββAbout: The About chapter has been reorganized, includes Bloomβs Taxonomy, and its outline was updated to reflect the addition of Chapter 4βββββChapter 1: Introduction: The file path was fixed, header formats were updated, and Bravoβs feedback was incorporatedβββββChapter 3: DL Primer: The file structure was reorganized, the chapterβs purpose was updated, and a giscus comment issue was resolvedβββββChapter 5: AI Workflow: The first draft of the AI workflow content was created and the overview, along with the purpose statement, were updatedβββββChapter 8: AI Training: The purpose of AI training and the explanation of training objectives were clarifiedβββββChapter 14: On-Device Learning: The purpose of on-device learning has been clarifiedβββββChapter 15: Security & Privacy: A duplicate case study was removed from the chapter, a duplicate section was moved to Chapter 14, and the chapterβs purpose was updatedβββββChapter 17: Sustainable AI: The chapterβs purpose was updated and incorporated feedback from BravoβββββAcknowledgements: The acknowledgements section now includes logos and an updated list of contributorsβββββSocratiQ: The SocratiQ chapter has been reorganized, broken links have been repaired, typos corrected, and an issue brought up by Bravo has been resolvedβββββChapter 6: Data Engineering: The purpose of the chapter was updatedβββββChapter 7: AI Frameworks: The purpose statement for Chapter 7 has been updatedβββββChapter 9: Efficient AI: The purpose statement for the chapter has been updatedβββββChapter 10: Model Optimizations: The purpose statement for Chapter 10 has been updatedβββββChapter 11: AI Acceleration: The purpose of Chapter 11 has been updatedβββββChapter 12: Benchmarking AI: The purpose statement for the chapter has been updatedβββββChapter 13: ML Operations: The purpose of Chapter 13 has been updatedβββββChapter 16: Responsible AI: The purpose of Chapter 16 has been updatedββββChapter 19: AI for Good: The chapterβs purpose has been updatedβββββChapter 20: Conclusion: The conclusion received some wording refinements and now includes a reference to Chapter 4
π§βπ» Labs
βββββLab: Raspi: The lab now includes a new VLM lab and the grid table formatting has been updatedβββββLab: Pi Vision Language Models: A new lab exploring Vision Language Models (VLMs) has been addedβββββLab: Labs Overview: Added VLM to the main table of labs
π 2024
November 19
π Chapters
βββββSocratiQ: The text was updated and an AI podcast was addedβββββAcknowledgements: The acknowledgements section now includes a list of contributors
November 19
π Chapters
βββββChapter 15: Security & Privacy: Machine unlearning was added to the chapter, a table was reordered, and a new federated learning case study replaced a previously discussed oneβββββChapter 16: Responsible AI: The chapterβs figure placement was refined, formatting was improved, policies were summarized, and a figure was corrected with a clearer explanationβββββChapter 17: Sustainable AI: The chapter now includes improved grammar, a new image illustrating the water footprint of AI, and an updated figure for Life Cycle AssessmentβββββChapter 19: AI for Good: The chapter now includes grammatical corrections and a discussion motivating the use of TinyMLβββββSocratiQ: Added .png support for gif images in PDF builds, updated the widget suggestions blog, and changed the variable name widget_access to socratiqβββββAcknowledgements: The acknowledgements section now includes a list of contributors
November 16
π Chapters
βββββChapter 16: Responsible AI: The explanation of a table has been improved for better clarityβββββAcknowledgements: The acknowledgements section now includes contributors and the preface material has been reorganizedβββββChapter 1: Introduction: A broken reference link was fixedβββββAbout: The fileβs content was reorganized and a relative path was corrected
November 15
π Chapters
βββββChapter 13: ML Operations: The chapter now features a more cohesive structure with grouped topics, removal of repeated content, clearer connections between roles, and a streamlined data management sectionβββββChapter 2: ML Systems: The Introduction section has been replaced with an OverviewβββββChapter 5: AI Workflow: The roles sections within Chapter 5 have been interconnected for a more cohesive flowβββββChapter 6: Data Engineering: The Introduction section has been replaced with an OverviewβββββChapter 16: Responsible AI: The chapter now features improved definitions and an overview section replacing the introductionβββββChapter 19: AI for Good: The Introduction section has been replaced with an OverviewβββββChapter 1: Introduction: Fixed definition formatting to be normal font and addressed style consistency errorsβββββChapter 3: DL Primer: The Introduction section has been replaced with an Overview sectionβββββChapter 7: AI Frameworks: The Introduction section has been replaced with an OverviewβββββChapter 8: AI Training: The Introduction section has been replaced with an Overview sectionβββββChapter 9: Efficient AI: The Introduction section has been replaced with an OverviewβββββChapter 10: Model Optimizations: The introduction section was replaced with an overviewβββββChapter 11: AI Acceleration: The introduction section was replaced with an overview sectionβββββChapter 14: On-Device Learning: The Introduction section was replaced with an Overview sectionβββββChapter 18: Robust AI: The Introduction section was replaced with an Overview sectionβββββChapter 15: Security & Privacy: The Introduction section has been replaced with an OverviewβββββChapter 20: Conclusion: The Introduction section has been replaced with an Overview sectionβββChapter 12: Benchmarking AI: The Introduction section has been replaced with an OverviewβββChapter 17: Sustainable AI: The Introduction section was replaced with an Overview section
π§βπ» Labs
βββββLab: Arduino Setup: The Introduction section has been replaced with an OverviewβββββLab: Arduino Motion Classification: The introduction section has been replaced with an overviewβββββLab: Raspberry Pi Setup: The Introduction section has been replaced with an OverviewβββββLab: Dsp Spectral Features Block: The introduction section has been replaced with an overview and the language has been simplified for improved readabilityβββββLab: Arduino Image Classification: The Introduction section was replaced with an Overview sectionβββββLab: Arduino Object Detection: The Introduction section has been replaced with an Overview sectionβββββLab: Arduino Keyword Spotting: The Introduction section has been replaced with an OverviewβββββLab: XIAO Setup: The Introduction section has been replaced with an Overview sectionβββββLab: XIAO Image Classification: The Introduction section has been replaced with an Overview sectionβββββLab: XIAO Object Detection: The Introduction section was replaced with an Overview sectionβββββLab: XIAO Keyword Spotting: The Introduction section has been replaced with an Overview sectionβββββLab: XIAO Motion Classification: The Introduction section has been replaced with an Overview sectionβββββLab: Pi Image Classification: The Introduction section has been replaced with an OverviewβββββLab: Pi Object Detection: The Introduction section has been replaced with an OverviewβββββLab: Pi Large Language Models: The Introduction section has been replaced with an Overview sectionβββββLab: Kws Feature Eng: The Introduction section was replaced with an Overview section
November 15
π Chapters
βββββChapter 1: Introduction: The definition of machine learning was updated, case studies were added with video and image links, and the definition of AI engineering was refinedβββββChapter 12: Benchmarking AI: The chapter now lacks a labs section, features reworked examples, and includes a new section discussing energy considerations within the context of benchmarkingβββββChapter 15: Security & Privacy: Added labs to the chapter, updated the security content, and renamed images to enhance privacy and securityβββββChapter 3: DL Primer: The DL Primer chapter was updated with new content and labs and core folders were createdβββββChapter 2: ML Systems: The chapter now includes labs and a core folder, features updated ML systems content, revised learning objectives, and an introduction that encompasses a broader scope than just embedded systemsβββββChapter 2: ML Systems: The labs portion has been removed from the chapters and moved into a new βlabsβ folder within the βcoreβ folderβββββChapter 3: DL Primer: The labs portion has been removed from the chapters and moved into a new βlabsβ folder, while the core content is now in a βcoreβ folderβββββChapter 5: AI Workflow: Labs have been removed from the chapters and placed in a dedicated βlabsβ folder, while the core content is now organized in a βcoreβ folderβββββChapter 6: Data Engineering: The labs portion has been removed from the chapters and moved into a new βlabsβ folder within the βcoreβ folderβββββChapter 7: AI Frameworks: Labs were removed from the chapter content and moved to a new βlabsβ folder, while a reference was addedβββββChapter 8: AI Training: The labs portion has been removed from the chapters and moved into a new βlabsβ folder within the βcoreβ folderβββββChapter 9: Efficient AI: The labs portion has been removed from the chapters and moved into a new βlabsβ folder within the βcoreβ folderβββββChapter 10: Model Optimizations: The lab exercises have been moved to a new βlabsβ folder within the existing βcoreβ folder structureβββββChapter 11: AI Acceleration: The labs portion has been removed from the chapters and moved into a dedicated βlabsβ folder within the βcoreβ folderβββββChapter 13: ML Operations: The ML Operations chapter now has a dedicated βlabsβ folder for practical exercises and incorporates feedback from BravoβββββChapter 14: On-Device Learning: The chapter content now focuses solely on theoretical concepts and explanations, removing the lab sections. Pruning is defined and discussed within the framework of on-device learningβββββChapter 18: Robust AI: The labs portion has been removed from the chapters and moved into a new βlabsβ folder within the βcoreβ folderβββββChapter 15: Security & Privacy: The chapter now excludes lab exercises and has been reorganized into βlabsβ and βcoreβ foldersβββββChapter 16: Responsible AI: The lab portion has been removed from the Responsible AI chapter and moved into a new βlabsβ folder within the βcoreβ folderβββββChapter 17: Sustainable AI: The labs portion has been removed from the chapters and moved into a dedicated βlabsβ folder within the βcoreβ folderβββββChapter 19: AI for Good: The labs portion has been removed from the chapters and moved into a new βlabsβ folder within the βcoreβ folderβββββAcknowledgements: The βacknowledgementsβ chapter now has dedicated βlabsβ and βcoreβ foldersβββββSocratiQ: A draft of the SocratiQ bot has been createdβββββSocratiQ: A draft of the SocratiQ bot and its QMD file have been createdβββββChapter 19: AI for Good: A labs folder was created and the AI4Good content was updatedβββββChapter 18: Robust AI: A labs folder was created and the robustAI content was updatedβββββChapter 17: Sustainable AI: A labs folder was created and the sustainable AI content was updatedβββββChapter 16: Responsible AI: A labs folder was created and the responsibleAI content was updatedβββββChapter 13: ML Operations: The ML Operations chapter now includes a labs and core folder structure and updated contentβββββChapter 14: On-Device Learning: The textbook now includes dedicated labs and core folders, and the on-device learning content has been updatedβββββChapter 12: Benchmarking AI: Labs and core folders were created and the benchmarking content was updatedβββββChapter 11: AI Acceleration: Labs and AI hardware content were added to the chapterβββββChapter 10: Model Optimizations: Labs and core folders were created, the βdevβ branch was merged, suggested fixes were implemented, and the optimization content was updatedβββββChapter 9: Efficient AI: Added labs and core folders and updated the efficiency contentβββββChapter 8: AI Training: A labs folder was created and the training content was updatedβββββChapter 7: AI Frameworks: The AI Frameworks chapter now includes dedicated labs and core folders, along with updated information on various machine learning frameworksβββββChapter 6: Data Engineering: The chapter now includes labs and a core folder, with updated data engineering content and an updated pathβββββChapter 5: AI Workflow: The AI Workflow chapter now includes dedicated labs and core folders, with an updated workflow overviewβββββChapter 1: Introduction: The introduction was updated with revised text and writing corrections, and new folders for labs and core content were created
π§βπ» Labs
βββββLab: Arduino Keyword Spotting: Fixed code examples, improved keyword spotting model accuracy, and added explanations for the model training processβββββLab: Labs: The lab content was restructured, documentation was improved, and the icon usage was adjusted to support PDF buildsβββββLab: Raspberry Pi Setup: The lab now includes notes on LLMs and other generic issues, incorporates Raspberry Pi support, and uses consistent formatting for emphasisβββββLab: Pi Image Classification: The lab now includes notes on LLMs and addresses other general issues, integrates Raspberry Pi support, and corrects the use of asterisksβββββLab: Pi Object Detection: Fixed the output format for object detection results and improved code comments for better understandingβββββLab: Pi Large Language Models: A lab focused on Small Language Models has been addedβββββLab: Labs Overview: The lab overview has been restructured to address issues related to PARTβββββLab: Nicla Vision: Corrected the use of colons in code examples for consistencyβββββLab: Xiao Esp32S3: Corrected the use of colons and asterisks in text elements for consistencyβββββLab: XIAO Setup: Corrected the use of colons and asterisks in text elements for consistencyβββββLab: XIAO Keyword Spotting: Corrected the use of colons in code examples for consistencyβββββLab: Raspi: Corrected punctuation in code examples and updated the content within the raspi.qmd file
January 02
π Chapters
βββββNiclav Sys: A typo was fixed in the Installing the OpenMV IDE sectionβββββChapter 7: AI Frameworks: The chapter content now includes corrections to a callout-tip and minor syntax errors
December 19
π Chapters
βββββChapter 10: Model Optimizations: Figures were added to the chapter and formatting issues along with broken references were corrected
December 18
π Chapters
βββββChapter 12: Benchmarking AI: The chapterβs content was moved to the benchmarks/leaderboards section and a problem with reference separators was correctedβββββChapter 17: Sustainable AI: Content related to sustainable AI was moved to the benchmarks/leaderboards section, wording about power draw was corrected, and the chapter was proofreadβββββChapter 10: Model Optimizations: Resolved a markdown formatting issue that affected Windows users
December 13
π Chapters
βββββChapter 7: AI Frameworks: The Colab notebooks for framework examples were updated
December 13
π Chapters
βββββChapter 9: Efficient AI: A broken URL link was fixed
December 12
π Chapters
βββββChapter 12: Benchmarking AI: Reference spacing in the chapter has been fixed
December 12
π Chapters
βββββChapter 8: AI Training: The file path for this chapter was updatedβββββChapter 10: Model Optimizations: A missing reference for an attention paper was added
December 12
π Chapters
βββββChapter 10: Model Optimizations: References now use square brackets instead of other formatting
December 11
π Chapters
βββββChapter 3: DL Primer: The DL primer no longer includes an activation function and its computation graph has been moved to the training sectionβββββChapter 8: AI Training: The DL primer activation function was removed and the computation graph was moved to the training sectionβββββChapter 12: Benchmarking AI: Several definitions were updated and a few examples were addedβββββGenerative Ai: Content in the chapter has been cleaned upβββββChapter 18: Robust AI: Text in the chapter has been cleaned upβββββChapter 6: Data Engineering: Various words were changed for clarity and precisionβββββEmbedded Ml: Various words were changed for clarity and precisionβββββEmbedded Sys: Various words were changed for clarity and precisionβββββChapter 11: AI Acceleration: Several words were changed for clarity and accuracyβββββChapter 14: On-Device Learning: Several words were changed for clarity and precision
December 11
π Chapters
βββββChapter 12: Benchmarking AI: The text now uses βTinyMLβ consistently throughout instead of βtinyMLββββββChapter 6: Data Engineering: The text now consistently uses βTinyMLβ instead of βtinyMLββββββEmbedded Sys: The text now consistently uses βTinyMLβ instead of βtinyMLββββββChapter 7: AI Frameworks: The text now uses βTinyMLβ consistently throughout instead of βtinyMLββββββChapter 11: AI Acceleration: The text now consistently uses βTinyMLβ instead of βtinyMLββββββLab: Arduino Image Classification: The text now consistently uses βTinyMLβ instead of βtinyMLββββββKws Nicla: The text now consistently uses βTinyMLβ instead of βtinyMLββββββChapter 10: Model Optimizations: The text now consistently uses βTinyMLβ instead of βtinyMLββββββChapter 16: Responsible AI: The text now consistently uses βTinyMLβ instead of βtinyMLββββββChapter 5: AI Workflow: The text now consistently uses βTinyMLβ instead of βtinyMLβ
December 11
π Chapters
βββββChapter 9: Efficient AI: References to datasets and ResNet architectures were added to the chapter, and a system for distributing references by chapter was implementedβββββChapter 7: AI Frameworks: A broken URL was fixed and references were distributed to each chapterβs own fileβββββChapter 10: Model Optimizations: The chapterβs terminology was updated for consistency, duplicate text about the lottery ticket hypothesis was removed, and references were moved to individual chapter filesβββββChapter 11: AI Acceleration: The chapter now includes references for machine learning and reinforcement learning applications in architecture design space exploration, general algorithm optimization, chip floorplanning, and logic synthesisβββββChapter 19: AI for Good: The AI for Good chapter now has its own dedicated references fileβββββChapter 12: Benchmarking AI: References are now distributed to individual chaptersβββββChapter 6: Data Engineering: Each chapter now has its own dedicated references fileβββββChapter 3: DL Primer: Each chapter now has its own references file, allowing for more specific and targeted citationsβββββDsp Spectral Features Block: Distributed references so each chapter has its own references fileβββββEmbedded Ml: Distributed references to individual chaptersβββββEmbedded Sys: Distributed references to individual chaptersβββββGenerative Ai: The Generative AI chapter now has its own dedicated references fileβββββLab: Arduino Image Classification: Each chapter now has its own dedicated references fileβββββKws Feature Eng: References are now distributed to individual chapter filesβββββKws Nicla: The Kws Nicla chapter now has a dedicated references fileβββββMotion Classify Ad: The Motion Classify Ad chapter now has its own dedicated references fileβββββNiclav Sys: The Niclav Sys chapter now has its own dedicated references fileβββββObject Detection Fomo: The Object Detection Fomo chapter now has its own dedicated references fileβββββChapter 14: On-Device Learning: Each chapter now has its own references file to improve organization and clarityβββββChapter 13: ML Operations: Distributed references so each chapter has its own references fileβββββChapter 15: Security & Privacy: Each chapter now has its own references file and the book was updated with changes from the βupstream/mainβ branchβββββChapter 16: Responsible AI: The Responsible AI chapter now has its own dedicated references fileβββββChapter 18: Robust AI: The Robust AI chapter now has its own dedicated references fileβββββChapter 17: Sustainable AI: The Sustainable AI chapter now has a dedicated references fileβββββChapter 8: AI Training: The AI Training chapter now has a dedicated references fileβββββChapter 5: AI Workflow: The workflow chapter now has its own dedicated references file
December 10
π Chapters
βββββChapter 8: AI Training: References now consistently appear before periods
December 10
π Chapters
βββββChapter 6: Data Engineering: The images folder was reorganized with subfolders by file type and all files were restructured into a new folder systemβββββDsp Spectral Features Block: The images folder was reorganized with subfolders based on filetype and all files were restructured into a new folder systemβββββEmbedded Ml: The image directory has been reorganized with subfolders by file type and all files have been restructured into a new folder systemβββββEmbedded Sys: The image directory was reorganized with subfolders based on file type and all files were restructured into a new folder systemβββββGenerative Ai: The image directory was reorganized with subfolders by file type and all files were restructured into a new folder systemβββββMotion Classify Ad: The image directory has been reorganized with subfolders based on file type and all files have been restructured into a new folder systemβββββNiclav Sys: The image directory has been reorganized with subfolders based on file type and all files have been restructured into a new folder systemβββββChapter 19: AI for Good: The image directory has been reorganized with subfolders based on file type and all files have been restructured into a new folder systemβββββChapter 12: Benchmarking AI: The image directory has been reorganized with subfolders based on file type and all files have been restructured into a more organized folder systemβββββChapter 3: DL Primer: The image directory has been reorganized with subfolders based on file type, and all files have been restructured into a new folder systemβββββChapter 9: Efficient AI: The image directory was reorganized with subfolders by file type and all files were restructured into a new folder systemβββββChapter 7: AI Frameworks: The image directory has been reorganized with subfolders by file type and all files have been restructured into a new folder systemβββββChapter 11: AI Acceleration: The image directory has been reorganized with subfolders based on file type, and all files have been restructured into a more organized folder systemβββββLab: Arduino Image Classification: The image classification lab now has subfolders within the images/ directory based on file type and all files have been reorganized into a more structured folder systemβββββKws Feature Eng: The images folder was reorganized with subfolders based on file type, and all files were restructured into a more organized folder systemβββββKws Nicla: The images folder was reorganized with subfolders based on file type and all files were reorganized into a more structured folder systemβββββObject Detection Fomo: The image directory was reorganized with subfolders based on file type and all files were restructured into a more organized folder systemβββββChapter 14: On-Device Learning: The images folder was reorganized with subfolders based on file type and all files were restructured into a new folder systemβββββChapter 13: ML Operations: The images folder was reorganized with subfolders based on file type and all files were restructured into a more organized folder systemβββββChapter 10: Model Optimizations: The image directory has been reorganized with subfolders based on file type and all files have been restructured into a new folder systemβββββChapter 15: Security & Privacy: The image folder has been reorganized with subfolders based on filetype and all files have been restructured into a new folder systemβββββChapter 16: Responsible AI: The image folder has been reorganized with subfolders based on filetype and all files have been restructured into a new folder systemβββββChapter 17: Sustainable AI: The image directory has been reorganized with subfolders based on file type, and all files have been restructured into a new folder systemβββββChapter 8: AI Training: The images folder is now organized by file type and all files have been reorganized into a new folder structureβββββChapter 5: AI Workflow: The images folder was reorganized with subfolders based on file type, and all files were restructured into a new folder systemβββββChapter 18: Robust AI: The files have been reorganized into a folder structure