CSE752: Deep Neural Networks

This course provides an in-depth exploration of deep learning. Topics include neural network basics, activation functions, forward and backward propagation; architectures like multi-layer perceptron (MLP) and fully connected layers; convolutional neural networks (CNNs) for vision tasks; recurrent neural networks (RNNs) including LSTMs and GRUs for sequential data; unsupervised learning methods like autoencoders, variational autoencoders (VAEs), and generative adversarial networks (GANs); transformer models such as BERT and GPT; optimization techniques including gradient descent variants, stochastic gradient descent (SGD), Adam, and learning rate schedules; regularization techniques like dropout and batch normalization; advanced architectures including ResNets and capsule networks; and transfer learning strategies for fine-tuning pre-trained models.

Course Objectives

The core objectives of this course are to:
To introduce fundamental concepts and architectures of deep learning.
To implement and train deep learning models using modern frameworks.
To provide an understanding of advanced architectures and optimization techniques.
To explore applications of deep learning in different domains.
To analyze the challenges and ethical considerations of deep learning models.

List of Books

1. To Be Added

Course Outcome