This repository provides a comprehensive curriculum for applied deep learning, targeting graduate students and advanced undergraduates with a strong foundation in statistics and linear algebra. It aims to familiarize learners with state-of-the-art industry techniques through a structured two-semester course, emphasizing practical coding skills in Python and frameworks like TensorFlow and PyTorch.

How It Works

The course is structured into two semesters, covering Computer Vision and Natural Language Processing in the first, and extending to Multimodal Learning, Generative Networks, and Reinforcement Learning in the second. It leverages a wealth of lecture notes, YouTube playlists, and links to seminal research papers with associated code repositories, offering a deep dive into foundational concepts and cutting-edge advancements.

Quick Start & Requirements

• Installation: No specific installation instructions are provided for the curriculum itself. Users are expected to set up their own Python environment and deep learning frameworks (TensorFlow/PyTorch).
• Prerequisites: Strong background in probability, statistics (linear & logistic regression), numerical linear algebra, and optimization. Proficiency in Python is required; familiarity with TensorFlow and PyTorch is beneficial.
• Resources: Access to YouTube and the ability to set up and run Python environments with deep learning libraries.

Highlighted Details

• Extensive coverage of both classic and recent deep learning architectures and techniques.
• Links to over 200 research papers, many with accompanying code implementations.
• Detailed breakdown of topics within Computer Vision (e.g., object detection, segmentation) and NLP (e.g., machine translation, language modeling).
• Includes advanced topics like Federated Learning, Self-Supervised Learning, and Graph Neural Networks.

Maintenance & Community

• The primary contributor is Maziar Raissi.
• The repository is linked to an arXiv preprint and a YouTube playlist, suggesting active development or a well-established course structure.

Licensing & Compatibility

• The repository itself does not specify a license. The linked research papers and code repositories will have their own licenses, which may vary.

Limitations & Caveats

• This is a curriculum and resource collection, not a runnable software package. Users must independently set up their development environment and execute code from linked repositories.
• The breadth of topics means that deep dives into specific areas might require consulting additional resources beyond those provided.