Machine Learning and Category Representation 2009-2010

Course at ENSIMAG, on Fridays between 9h-12h, in the ENSIMAG building in Montbonnot. Lectures by Cordelia Schmid and Jakob Verbeek.

Session 1, September 25 2009

Introduction (Cordelia Schmid) slides.

Session 2, November 27 2009

Object Recognition: History and overview, bag of features classification (Cordelia Schmid) slides.
Introduction machine learning (Jakob Verbeek) slides.

Session 3, December 4 2009

Clustering with k-means, mixture of Gaussians (Jakob Verbeek) slides.
Local invariant features (Cordelia Schmid) slides.

Session 4, December 11 2009

Local Features (Cordelia Schmid). slides.
Mixture of Gaussians, EM algorithm continued (Jakob Verbeek) slides.
Reading material:
- A View of the EM Algorithm that Justifies Incremental, Sparse, and Other Variants
- extract Chapter 3 thesis J. Verbeek

Session 5, December 18 2009

Classification 1: generative and non-parameteric methods + Fisher kernels (Jakob Verbeek) slides.
Reading material:
- Fisher Kernels on Visual Vocabularies for Image Categorization by Perronnin and Dance, CVPR'07.
Visual Search (Cordelia Schmid) slides.

Session 6, January 8 2010

NOTE: lecture postposed to Monday January 11 due to the snow
Classification 2: discriminative methods and kernels (Jakob Verbeek).
Topic models (Jakob Verbeek) slides.

Session 7, January 15 2010

Category level localization: Sliding window and shape model (Cordelia Schmid). slides.

Exam

9h-12h, Friday January 29, 2010. Room D.207 (ensimag).
prepare from lecture slides, papers given as reading material, and last year's exam.

Form: closed book, i.e. it is not allowed to bring any material.