IFT 6132 : Lectures - Winter 2024

Books on structured prediction

Part I: theory of structured prediction

Lecture 1 – 1/9 – introduction

Lecture 2 – 1/11 – structured prediction elements and examples

Lecture 3 – 1/16 – Energy-based models

Lecture 4 – 1/18 – Theory (binary classification)

Lecture 5 – 1/23 – Theory (structured prediction) - PAC-Bayes

Lecture 6 – 1/25 – Theory: generalization error bounds

Lecture 7 – 1/30 – Kernels and RKHS

Lecture 8 – 2/1 – Theory: calibrated convex surrogate losses

(No lectures on 2/6 and 2/8)

Part II: convex optimization and structured prediction

Lecture 9 – 2/13 – convex optimization

Lecture 10 – 2/15 – stochastic subgradient method

Lecture 11 – 2/20 – convex optimization (II) + structured SVM optimization (I)

Lecture 12 – 2/22 – structured SVM optimization (II) - small QP

Lecture 13 – 2/27 – structured SVM optimization (III) - M3-net

Lecture 14 – 2/29 – structured SVM optimization (IV) - duality

Week of 3/4 - spring break (no lecture on 3/5 and 3/7)

3/12 – lecture cancelled

Lecture 15 – 3/14 – structured SVM optimization (V) - M3-net dual & cutting plane

Lecture 16 – 3/19 – FW & AFW algorithms

Lecture 17 – 3/21 – FW convergence

Lecture 18 - 3/26 – FW for SVMstruct

Lecture 19 - 3/28 – SAG method

Lecture 20 - 4/2 – CRF and variance reduced SGD

Lecture 21 – 4/4 – catalyst

lecture 22 – 4/9 – latent variable; RNN

lecture 23 – 4/11 – learning to search; SeaRNN; SPENs

Poster session – 4/30