EECS 497 - Advanced Topics in Computational Learning Theory
This course will cover advanced topics in computational learning theory.
Students will present papers and discuss open problems.
The prerequisite is EECS 496-10: Computational Learning Theory,
or equivalent preparation.
In addition to attending lectures,
each registered student is expected to present two papers (approximated for 30 minutes each).
Time and Location: W 9:30am-12:20pm, Tech M166
Lecture 1 (4/3/19)
presentation by instructor Efficient Learning of Typical Finite Automata from Random by Freund et al.
Lecture 2 (4/10/19)
presentation by instructor Learning Finite Automata Using Label by Angluin et al. and
Open Problem: Meeting Times for Learning Random by Fish and Reyzin
Lecture 3 (4/17/19)
presentation by Liren Shan Linear Contextual Bandits with (on his joint work with Lou and Shan)
presentation by Aravind Reddy on "Understanding deep learning requires rethinking generalization" by Zhang et al.
Lecture 4 (4/24/19)
presentation by Thomas On the Sample Complexity of the Linear Quadratic by Dean et al.
presentation by Abhratanu On learning mixtures of well-separated by Regev and Vijayaraghavan
Lecture 5 (5/1/19)
Lecture 6 (5/8/19)
presentation by Aidan Smoothed Analysis of Discrete Tensor Decomposition and Assemblies of by Anari et al.
presentation by Jason Hartline Econometrics for Learning by Nekipelov et al.
Lecture 7 (5/15/19)
presentation by Abhratanu Dutta Computational Cost for Achieving Adversarial Robustness for (his joint work with Awasthi and Vijayaraghavan)
Lecture 8 (5/22/19)
presentation by Aravind A Theoretical Analysis of Contrastive Unsupervised Representation by Arora et al.
Correlation Clustering with local (on his work with Makarychev and Zhou)
Lecture 9 (5/29/19)
Aidan Efficient Profile Maximum Likelihood for Universal Symmetric Property
Thomas Adversarial Examples Are Not Bugs, They Are by Ilyas et al.
Lecture 10 (6/5/19)
Sanchit Algorithms and Hardness for Robust Subspace by Hardt and Moitra
talk by instructor on differential privacy and adaptive data analysis