DS-UA 301 Advanced Techniques in Machine Learning and Deep Learning

DS-UA 301 Advanced Topics in Data Science:
Advanced Techniques in Machine Learning and Deep Learning

Term: Fall 2022
Instructor: Dr. Parijat Dube
Level: Undergraduate

Topics

AI revolution, ML concepts, Introduction to ML tooling; Introduction to Deep Learning (DL);

DL Training Tools and Techniques; Special Deep Learning Architectures; Transfer Learning and Semi-supervised Learning;

Hyperparameter Optimization and Feature Engineering; Automated Machine Learning; Robust Machine Learning;

Distributed Training and Federated Learning; Model drift and Continual learning;

Description

This course provides a practical introduction to tools and techniques for different stages of machine learning life-cycle from model development to post-deployment. We will study techniques for neural network training, hyperparameter optimization, feature engineering, and neural architecture search. Students will gain knowledge of automated ML tools like auto- sklearn, auto-WEKA, H20.ai. Beyond model development, the class will be introduced to tools for creating robust models, explaining model predictions, conducting bias and fairness audits, detecting post-deployment model drift and continual training. Emphasis will be on getting working knowledge of tools and techniques through extensive lab work and hands-on assignments involving standard DL frameworks (Tensorflow, Pytorch) and open-source technologies.