3 Credit Hour Course
Intended For Level 0 Term 0 Students
Prerequisite:
Foundations of Neural Networks and Deep Learning: components of a learning algorithm, activation functions, loss functions, back propagation, multi-layer perceptron, regularization, dropouts, weight decay, batch normalization, optimization algorithms; Convolutional Neural Networks (CNN): convolution and pooling, variants of convolutional layers, dilated convolution, transfer learning; Recurrent Neural Networks (RNN): computing gradients in RNN, deep RNN; sequence-to-sequence architectures, word embedding, recursive networks, backpropagation through time, vanishing and exploding gradients, long short term memory (LSTM), self attention, transformer; Deep Unsupervised Learning: autoencoders, variational autoencoders, generative adversarial networks; Advance topics: graph neural networks, deep reinforcement learning, attention and memory models.
