T81-558 is an educational course repository that teaches deep learning applications using Keras and TensorFlow through a combination of lecture materials, Jupyter notebooks, and practical assignments at Washington University in St. Louis.
The course addresses the problem of learning how to apply deep learning techniques to real-world problems across multiple domains. It covers classical neural network structures alongside modern architectures including convolutional neural networks, long short-term memory networks, gated recurrent networks, and generative adversarial networks. The approach combines theoretical foundations with hands-on implementation using Python, TensorFlow, and Keras, with applications spanning computer vision, time series analysis, natural language processing, audio processing, and reinforcement learning. The material also covers high-performance computing aspects for leveraging GPUs and distributed computing grids.
This repository is designed for students entering a formal university course, though the complete textbook is available publicly on GitHub. The material assumes familiarity with at least one programming language but does not require prior Python knowledge. The course is structured as a hybrid delivery combining classroom and online instruction. Current students at Washington University should note that the university has transitioned to a PyTorch version of this course, making this repository the previous generation of the offering. The repository is most suitable for learners seeking a comprehensive introduction to deep learning applications rather than those looking for cutting-edge research implementations.
The project maintains active engagement with its issue tracker, where work is dominated by bug reports and enhancement requests, indicating ongoing refinement of the course materials and code examples.
