Information Theory in CS

Swedish Summer School in Computer Science,

KTH, Stockholm, July 1-5, 2019

Lecturer: Madhu Sudan

Lectures:

Note: Problem sets corresponding to some of the lectures are available on request. Please email Madhu if you wish to use those in some course you are teaching.

  • Day 1:
    • Topics:
      • Part 1: Shearer's Lemma/Loomis-Whitney Inequality; Informal definition of entropy + axioms; Implication: LWI/Shearer.
      • Part 2: Formal definition of entropy; proof of axioms; divergence theorem; divergence; other stuff. Reading
      • Part 3: Entropy vs/ Compression: single-shot; multi-shot; prefix-free compression;
    • Reading(s): Some subset of L1-1.pdf, L1-2.pdf, L1-3.pdf
  • Day 2:
    • Topics:
      • Part 1: Kraft's inequality; Shannon compression; Huffman compression. Markovian Sources. Universal Compression.
      • Parts 2 & 3: Lempel-Ziv. Claims. Parts of Proofs.
    • Reading(s): Some subset of L2-1.pdf, L2-2.pdf, L2-3.pdf and L2-4.pdf. (Addendum: See also this paper with David Xiang for a full analysis of LZ compression.)
  • Day 3:
    • Topics:
      • Part 1: Error-correction. Binary symmetric channel. Capacity and (speed of) convergence to  capacity.
      • Part 2: Polar codes. Martingales and polarization theorem. Encoding, Decoding.
      • Part 3: Strong Polarization of the polar-code-martingale.
    • Reading(s): Parts of L2-3.pdf, L2-4.pdf, L3-1.pdf, L3-2.pdf and L3-3.pdf.
  • Day 4:
    • Topics:
      • Part 1: Introduction to Communication Complexity.
      • Parts 2 & 3: Information Complexity and Lower bound on Disjointness.
    • Reading(s): Parts of L4-0.pdf, L4-1.pdf, L4-2.pdf.
  • Day 5:
    • Topics:
      • Part 1: Compressing Interactive Communication
      • Part 2: Information = Amortized Communication
      • Part 3: Parallel Repetition Theorem.
      • Readings: Parts of
    • Readings: Parts of L4-4.pdf, L4-5.pdf, L4-6.pdf, L5-2.pdf, L5-3.pdf