6.883 Supervised Learning: Bridging Theory and Practice

SHARE:

Graduate Level
Units:  3-0-9
Prereqs: 6.046 or equivalent; 6.042/6.041/6.008 or equivalent; 6.036 or equivalent
Instructor:  Professor Constantinos Daskalakis (costis@csail.mit.edu)
Schedule:  MW2:30-4, room 24-115
 
Description:
This subject qualifies as an Artificial Intelligence concentration subject.
 
Recent advances in Deep Learning have enabled tremendous progress in learning and generating high-dimensional complex structures. But what do deep methods really learn? And how can we decrease their sample complexity or supervision? 
This class, situated at the intersection of machine learning and high-dimensional statistics, will take a principled approach to these questions through a combined theoretical and practical approach. Topics include Generative Adversarial Networks (GANs), Graphical Models, High-dimensional learning and testing, Causality, Interpretability, strategic behavior and adversarial training.