Deep Learning Theory And Applications

Catalog description

This course will cover both the practical and theoretical foundations of deep neural networks (DNN). Key topics include probabilistic foundation of DNN, computation graphs, DNN architectures and optimization methodologies. In this course, we aim to a wide range of deep neural network architectures for different applications, from basic models to advanced ones like auto-encoders, convolutional networks, recurrent neural networks (e.g. transformers) and graph neural networks and large language models. Additionally, this course discusses deep generative models such as, variational auto-encoders, generative adversarial networks. Real-world applications (e.g., image segmentation and classification, image and audio generation) from different fields will be showcased throughout the course. The tutorials will enhance understanding by providing hands-on experience in implementing and using deep neural networks with Keras and PyTorch. Prerequisite(s): ECE 31100 FOR LEVEL UG WITH MIN. GRADE OF D- AND ECE 49500 FOR LEVEL UG WITH MIN. GRADE OF D- Course Learning Outcomes 1. Understand Fundamentals of Deep Learning: Key concepts such as neural networks, computation graphs, activation functions, loss functions, and optimization techniques. 2. Implement Fundamental and Advanced Deep Learning Models in Python: Build, train, and evaluate various deep learning models using Python and deep learning libraries such as PyTorch and Keras. 3. Optimize Deep Learning Models: Understand and apply regularization techniques, hyperparameter tuning, and other optimization methods to improve the generalization and performance of models. 4. Analyze and Apply Deep Neural Network Architectures: Practical experience with advanced architectures, including regular and advanced architectures from different categories of deep learning architectures including convolutional neural networks, graph neural networks, recurrent neural networks, large language models. 5. Understand and Apply Generative Models on Real-world Applications: Explore generative modeling vs. discriminative modeling and implement deep generative models like, variational autoencoders, flow-based models, generative adversarial networks, and large language models. View Class Schedule