Deep Learning for Wireless Communication

Course Description

This course provides a fundamental understanding of neural networks (NNs) with a focus on applications related to wireless signals. The course will cover different NNs, activation functions and intuition to create application specific loss functions. Topics include Convolutional Neural Network (CNN), Recurrent Neural Networks (RNN), Autoencoders (AE) and Generative Adversarial Networks (GAN) with examples from wireless applications, like waveform generation, channel coding, modulation detection, channel estimation, RF fingerprinting and wireless security. As NNs are mostly developed and tested with images, many concepts cannot be borrowed in the wireless domain due to the fundamental differences in data type. This course will take a gray box approach for understanding the NNs to help bridge the gap between wireless signal processing and deep learning.

Books

Highly Recommended:

"Dive into Deep Learning", interactive book - https://d2l.ai
Ian Goodfellow and Yoshua Bengio and Aaron Courville, "Deep Learning", https://www.deeplearningbook.org.
Kevin P. Murphy, "Machine Learning: A Probabilistic Perspective", https://probml.github.io/pml-book/.
Recent papers on Deep Learning based Wireless Communication

Course Logistics

This course will be delivered completely online over Webex.
Course Dates: Check schedule below.
Course Times: 12-2PM

Course Calendar

Subject to change as per travel schedule, material progreess and other unforseen events

Week	Date	Discussion Topic	Other Information
1	Jul 7	Introduction, Machine Learning Basics	Complete Slide Deck
1	Jul 9	Linear Neural Networks, Gradient Descent, Hyperparameters
2	Jul 12	Classification, Softmax, Cross-entropy
2	Jul 14	Multilayer Perceptron and non-linear activation functions
3	Jul 19	Forward and Backward Propagation, Regularization
3	Jul 21	CNN, Stride, Padding, Pooling
4	Jul 26	Sequence Modeling, RNN, Backpropagation through time, GRU
4	Jul 28	Autoencoder, GAN, Federated Learning, Transformer NN