# notes

### Technical Reviews

- What Can ResNets Learn Efficiently? [pdf] (for Simons Summer 2019 Reading Group on Generalization Theory)
- Geometry of Optimal Transport [pdf] (for Simons Summer 2019 Reading Group on Optimal Transport)
- Learning to Plan and Hierarchical, Multi-Task Reinforcement Learning [pdf] (for IEOR 8100)
- Generalization from Margins: Connecting Boosting Theory to Neural Nets [pdf–in progress!] (at Columbia DSI Seminar)
- Learning Sparse Polynomials [pdf] (for COMS 6998-5)
- Interactive Clustering [pdf] (for COMS 6998-4)
- Convex Optimization and Randomness [pdf] (for Dave Blei’s Reading Group)
- Tensor Decompositions [pdf] (for ELE 538B)
- Rate Distortion and Unsupervised Learning [pdf] (for COS 598E)
- Matrix Concentration and Applications [pdf] (for ORF 550)
- On Lipschitz Extensions from Finite Sets [pdf] (with Mikhail Khodak, for MAT 529)
- The Structure Theorem for Finitely-Generated Modules over PIDs [pdf] (for MAT 346: Algebra II)
- The Representation of Language in the Brain [pdf] (at Algorithms-Machine Learning Reading Group)
- A Brief Survey on Expander Graphs [pdf] (for Junior Seminar with Zeev Dvir)

### Class Notes

For now, I have individual class notes for some classes I’ve taken. Sometimes, lectures are missing or the rest of the notes have been taken by other people — in these events, I direct you to the website of the course, if it exists. If not, take a look at my links page for a list of other resources. At some point in the future, I plan to put together a big document of notes organized by subject.

- COMS 6998-9: Algorithms for Massive Data [pdf] [course website]
- COMS 4995-2: Unsupervised Learning [pdf] [course website]
- STAT 8301: Topics in Probability Theory (taught by Roman Vershynin) [pdf–in progress!] [all notes]
- IEOR 8100: Reinforcement Learning [pdf] [course website]
- STAT 8101: High-dimensional Statistical Inference [pdf–in progress!] [course website]
- COMS 6998-5: Algorithms with a Geometric Lens [scribe notes] [course website]
- COMS 6998-4: Interactive Learning [scribe notes] [course website]
- COS 598E: Unsupervised Learning [scribe notes] [course website]
- ORF 550: Probability in High Dimension [scribe notes] [syllabus] [all notes] (with Xinyi Chen)
- ORF 524: Statistical Theory and Methods [scribe notes] [syllabus]
- MAT 529: Metric Embeddings and Geometric Inequalities (taught by Assaf Naor) [pdf] (with Holden Lee)
- MAT 597/PHY521: Mathematical Physics (taught by Michael Aizenman) [pdf] (with Holden Lee)
- APC 529: Coding Theory and Random Graphs (taught by Emmanuel Abbe) [pdf]
- MAT 340: Applied Algebra (taught by Mark McConnell) [pdf]
- CS 224d: Deep Learning for Natural Language Processing (taught by Richard Socher) [scribe notes]
- COS 511: Theoretical Machine Learning [scribe notes] (with others)
- COS 510: Programming Languages (taught by David Walker) [scribe notes]
- APC 486: Transmission and Compression of Information (taught by Emmanuel Abbe) [scribe notes]

### Seminar Notes

- Yann LeCun on
*The Power and Limits of Deep Learning and AI*[pdf] - Yisong Yue on
*New Frontiers in Imitation Learning*[pdf] - Tim Roughgarden on
*How does Computer Science Inform Modern Auction Theory*[pdf] - Yuchen Zhang on
*Two Approaches to Non-Convex Machine Learning*[pdf] - Ben Recht on
*What can Deep Learning learn from Linear Regression*[pdf] - Nicholas Boumal on
*ManOpt: Manifold Optimization*[pdf] - Michael Jordan on
*Computational Thinking, Inferential Thinking, and Data Science*[pdf] - Joel Tropp on
*Universality Laws for Randomized Dimension Reduction*[pdf] - Joel Tropp on
*Finding Structure With Randomness*[pdf] - Michael Jordan on
*Nonparametric Bayesian and Combinatorial Stochastic Processes*[pdf] - Jacob Steinhardt on
*Unsupervised Risk Estimation with Only Structural Assumptions*[pdf] - Yuanzhi Li on
*Weighted Low-Rank Matrix Approximation*[pdf] - Zeyuan Allen-Zhu on
*Linear Coupling Framework of Gradient and Mirror Descent*[pdf] - Naman Agarwal and Brian Bullins on
*LiSSA: A Linear Time Second-Order Stochastic Algorithm*[pdf] - Andrew Barron on
*Computationally Feasible Greedy Algorithms for Neural Networks*[pdf] - Samy Bengio on
*Neural Image Captioning*[pdf] - Zaid Harchaoui on
*Convolutional Kernel Neural Networks*[pdf] - Lester Mackey on
*Divide-and-Conquer Matrix Completion*[pdf] - Gillat Kol on
*Interactive Information Theory*[pdf] - Sanjeev Arora on
*Reversible Deep Nets*[pdf] - Santosh Vempala on
*The Complexity of Detecting Planted Solutions*[pdf] - Amir Ali Ahmadi on
*Optimizing over Nonnegative Polynomials*[pdf] - Francisco Pereira on
*Decoding Generic Representations of fMRI*[pdf] - Elad Hazan on
*Simulated Annealing and Interior Point Methods*[pdf] - Dana Lahat on
*Joint Independent Subspace Analysis and Blind Source Separation*[pdf] - Barbara Engelhardt on
*Bayesian Structured Sparsity Using Gaussian Fields*[pdf] - Dimitris Bertsimas on
*Statistics and Machine Learning from a Modern Optimization Lens*[pdf] - Sébastien Bubeck on
*Optimal Regret Bounds for the General Convex Multi-Armed Bandit Setting*[pdf] - Han Liu on
*Nonparametric Graphical Models*[pdf] - Mehryar Mohri on
*Deep Boosting*[pdf] - Percy Liang on
*Learning Hidden Computational Processes*[pdf] - Young Kun Ko on
*The Hardness of Sparse PCA*[pdf] - Tom Griffiths on
*Rationality, Heuristics, and the Cost of Computation*[pdf] - Anna Choromanska on
*Optimization for Large-Scale Machine Learning*[pdf]

### Symposia Notes

- MIT Deep Learning and Nonconvex Optimization (MIFODS MIT Workshop) (Jan. 2019) [page]
- MIT Sublinear Algorithms, Local Algorithms, and Robust Statistics (MIFODS MIT Workshop) (Jun. 2018) [page]
- Bridging Optimization Theory, Information Theory, and Data Science at Princeton (May. 2018) [pdf]
- Mathematics of Deep Learning at Princeton (Mar. 2018) [pdf]
- NYU Mini-Theory Symposium (Feb. 2018) [pdf]