This repository provides the Keras/TensorFlow implementation for the "Applications of Deep Neural Networks" course (T81-558) at Washington University in St. Louis. It's designed for students and practitioners seeking to understand and apply various deep learning architectures like CNNs, LSTMs, GANs, and reinforcement learning to diverse data types including tabular data, images, and text. The course material, including a textbook, is available on GitHub.

How It Works

The course material is structured into modules covering Python fundamentals for machine learning, TensorFlow/Keras basics, and advanced deep learning topics. It emphasizes practical application, demonstrating how to implement and train neural networks for tasks such as classification, regression, computer vision, natural language processing, and time series analysis. The approach focuses on using Python with TensorFlow and Keras, with some modules delving into high-performance computing aspects using GPUs.

Quick Start & Requirements

• Installation: Primarily involves setting up a Python environment with TensorFlow and Keras. Specific commands depend on the user's setup.
• Prerequisites: Python 3.x, TensorFlow, Keras, Pandas, Scikit-learn. GPU and CUDA are beneficial for performance but not strictly required for all examples.
• Resources: Access to datasets is provided via a link. Setup time is dependent on user's Python environment familiarity.
• Links: Course Textbook, Syllabus, Datasets

Highlighted Details

• Covers a broad range of deep learning architectures: CNNs, LSTMs, GRUs, GANs, and Reinforcement Learning.
• Includes practical applications in computer vision, time series, NLP, and data generation.
• Demonstrates techniques like transfer learning, regularization, dropout, and hyperparameter optimization.
• Features modules on using pre-trained models and integrating with platforms like Hugging Face.

Maintenance & Community

The repository is maintained by Jeff Heaton, the course instructor. The README encourages following him on GitHub for updates. There are no explicit links to community forums like Discord or Slack provided.

Licensing & Compatibility

The repository content appears to be freely available for educational purposes. Specific licensing details are not explicitly stated in the README, but the material is presented as a course offering.

Limitations & Caveats

This repository contains the previous Keras/TensorFlow version of the course; the current university offering uses PyTorch. While Python familiarity is assumed, prior programming experience is recommended. The course focuses on applications, with mathematical foundations introduced rather than deeply explored.