This repository provides resources for the "Machine Learning for the Web" class at ITP, NYU, focusing on creative applications of ML in the browser. It's designed for students with JavaScript and HTML/CSS experience, aiming to equip them with practical skills in using and training ML models for interactive web experiences, covering topics from image classification to text generation.

How It Works

The course leverages high-level JavaScript libraries like ml5.js and TensorFlow.js to enable browser-based ML model execution and training. It progresses from using pre-trained models to building custom models from scratch, emphasizing creative outputs and practical application over deep theoretical understanding. The curriculum covers techniques such as Transfer Learning, CNNs, GANs, and latent diffusion.

Quick Start & Requirements

• Clone the repository: git clone https://github.com/yining1023/machine-learning-for-the-web.git
• Navigate and serve: cd machine-learning-for-the-web then python3 -m http.server
• Access examples via localhost:8000.
• Prerequisites: JavaScript, HTML, CSS proficiency (equivalent to an ICM course). A modern laptop (under 4 years old) is recommended.

Highlighted Details

• Covers a wide range of ML applications: Image/Sound/Doodle Classification, Face/Pose/Hand Recognition, Image, Video, and Text Generation.
• Explores various ML techniques: Transfer Learning, KNN Classifier, PoseNet, BodyPix, U-Net, Face-api, Facemesh, Handpose, Word2Vec, CharRNN, and GANs.
• Includes practical coding sessions and assignments for each topic.
• Provides extensive resources for ml5.js, TensorFlow.js, and related ML concepts.

Maintenance & Community

The repository is associated with ITP, NYU, and instructor Yining Shi. Specific community channels or active maintenance beyond the course context are not detailed in the README.

Licensing & Compatibility

The README mentions checking licenses for code usage and fulfilling their terms, referencing choosealicense.com. It strongly encourages citing sources and using libraries, with a warning against plagiarism. Specific repository-wide licensing is not explicitly stated.

Limitations & Caveats

This repository serves as course material and may not represent a production-ready framework. The focus is on educational application, and the content reflects a specific course structure and timeline (Jan-May 2023).