Info
Home
Homework
Class notes
Paper reading
Data project

ECE 598IS - Fundamental Limits in Data Science (Spring 2022)

Scribe:
  • Each student will be expected to be a scribe for at least one lecture.
  • You can sign up for a lecture here
  • Scribe Latex template
  • The scribe is due one week after the lecture

Lecture notes

  • Handwritten lecture notes can be found here.
  • Lecture 1 (01/18). Introduction (video)
  • Lecture 2 (01/20). Stastical decision theory (scribe) (video)
  • Lecture 3 (01/25). From Estimation to Testing (scribe)
  • Lecture 4 (01/27). Divergence measures (total variation and KL divergence) (scribe)
  • Lecture 5 (02/01). Squared Hellinger distance and Le Cam's two-point method (scribe)
  • Lecture 6 (02/03). Information measures and inequalities (scribe) (video)
  • Lecture 7 (02/08). Fano's Method and distance-based Fano's Method (scribe)
  • Lecture 8 (02/10). k-sparse Gaussian location model (scribe)
  • Lecture 9 (02/15). Lower bound for compressive sensing (scribe)
  • Lecture 10 (02/17). Finished compressive sensing. The planted clique problem. (scribe)
  • Lecture 11 (02/22). Linear Algebra review (scribe)
  • Lecture 12 (02/24). Eigenvalue and Eigenvector perturbations (scribe)
  • Lecture 13 (03/01). Operator norm of a random matrix (scribe)
  • Lecture 14 (03/03). Spectral Estimator for Planted Clique (scribe)
  • Lecture 15 (03/08). Planted Partition Problem (scribe)
  • Lecture 16 (03/10). Stochastic Block Model Exact Recovery (scribe)
  • Lecture 17 (03/22). Spectral Estimator for SBM (scribe)
  • Lecture 18 (03/24). Semidefinite Programming Estimator for SBM (scribe)
  • Lecture 19 (03/29). "Information Limits for Recovering a Hidden Community" (slides) and "Community Recovery in Graphs with Locality" (slides)
  • Lecture 20 (03/31). "Optimal Sampling and Clustering in the Stochastic Block Model" (slides) and "Information-theoretic limits of selecting binary graphical models in high dimensions" (slides)
  • Lecture 21 (04/05). "On the Power of Louvain in the Stochastic Block Model" (slides) and "Stochastic Block Model for Hypergraphs: Statistical limits and a semidefinite programming approach"
  • Lecture 22 (04/07). Sequence reconstruction and the trace reconstruction problem (scribe)
  • Lecture 23 (04/12). Lower bound for trace reconstruction (scribe) + "Trace reconstruction with exp(O(n1/3)) samples" (slides) + "Coded trace reconstruction" (slides)
  • Lecture 24 (04/14). Sequence reconstruction from substrings (scribe) + "Optimal Assembly for High Throughput Shotgun Sequencing" (slides)
  • Lecture 25 (04/19). Dimensionality reduction, Johnson-Lindenstrauss, PCA (scribe)
  • Lecture 26 (04/21). PCA on estimated covariance matrix (scribe)
  • Lecture 27 (04/26). Sparse PCA (scribe)
  • Lecture 28 (04/28). Sketching and Locality Sensitive Hashing (scribe)
  • Lecture 29 (05/03). "Iterative Hessian sketch: Fast and accurate solution approximation for constrained least-squares" (part 1, part 2) and "Adaptive Learning of Rank-One Models for Efficient Pairwise Sequence Alignment" (slides)