EPFL Spring Semester 2013
EPFL Spring Semester 2013
This course (EE-612) presents fundamental tools used in statistical pattern recognition ranging from the most basic (LR, PCA, LDA, MLP, GMM, HMM) to the some more elaborated (ISV, JFA, PLDA). This course could serve as a pre-requisite for more advanced course on Machine Learning.
The main instructor is Sébastien Marcel, and Labs will be shared with André Anjos, Laurent El-Shafey and BOB.
Program:
• Lectures: 36 hours (9 lectures of 4 hours)
• Labs: 20 hours (4 labs of 5 hours)
• Exam form: project + presentation
• Room: ELG 116
Required prior knowledge:
Linear algebra, Probabilities and Statistics, Signal Processing
About the slides and the labs:
In general the slides of a lecture are available after the lecture (later in the day), while the lab material is available before the lecture (early morning).
Content:
• Lecture 0: Overview -- Feb 21 2013
• Lecture 1: Introduction -- Feb 21 2013
• Lecture 2: Linear Regression (univariate) -- Feb 28 2013
• Lecture 3: Linear Regression (multivariate) -- Mar 7 2013
• Lab 1: Linear Regression with Python -- Mar 14 2013
• Lecture 4: Logistic Regression -- Mar 21 2013
• Lab 2: Logistic Regression with Python and BOB -- Mar 28 2013
• Lecture 5: Multi-Layered Perceptron (1/2) -- Apr 11 2013
• Lecture 6: Multi-Layered Perceptron (2/2) -- Apr 18 2013
• Lab 3: Multi-Layered Perceptron -- Apr 25 2013
• Lecture 7: Dimensionality reduction and Clustering -- May 2 2013
Warning !! This lecture will be given in room BC02 not ELG116
• Lecture 8: Probability Distribution Modeling -- May 16 2013
In particular Gaussian Mixture Models and Expectation-Maximization !
• Lecture 9: Session Variability Modeling (advanced lecture) -- May 23 2013
Inter-Session Variability modeling (ISV), Joint Factor Analysis (JFA), Total Variability modeling (TV) aka iVectors, and Probabilistic Discriminant Analysis
• Lab 4: PCA, LDA, k-Means and GMMs -- May 30 2013
• Exam: Homework project given on May 23 1013 to complete (short report + source code) by June 13 (11:59pm) and to present on June 17 in room CO 121.
Labs material:
• BOB: http://idiap.github.com/bob/
• BOB short course ( https://github.com/idiap/bob/wiki/Bob-Starter-Course ): This is an introductory course on Python, Bob and Scientific Software Management.
• A YouTube playlist is available HERE for this short course on BOB
Fundamentals in Statistical Pattern Recognition