Class Schedule:
Lectures : M T Th at 11am in WL228
Exam Schedule:
MidSem - September 21
Numerical Assignment - October ?
EndSem - November ?
Books/References (with links):
A Probabilistic Theory of Pattern Recognition by Luc Devroye, László Györfi and Gábor Lugosi.
The Elements of Statistical Learning: Data Mining, Inference, and Prediction by Trevor Hastie, Robert Tibshirani, Jerome Friedman.
Miscellaneous:
Pattern Classification by Richard Duda, Peter Hart and David Stork , John Wiley & Sons, 2001.
Multivariate Density Estimation by David W. Scott (available in IITK library)
Some Useful Links (Glossary): 1 , 2 , 3
Notes - Elliptic symmetry , Finiteness of mgf , Order Statistics ,
Corrections (old) - Logistic regression , Cover of $\mathbb{R}^d$ - Thanks again to Rahul Singh! ,
Solutions (old) - Quiz 1 , Midsem Q. 2(c) - Thanks to Rahul Singh! ,
Topics:
Parzen Window video : The Parzen window estimate of a pdf (thin black line) matches with the actual pdf (thicker blue line). The histogram of the actual data points are shown in light gray in the background.
M-estimation and this (from empirical process theory)
SVM : 1 , 2 and Papers - 1 , 2 , 3
Curse (Blessing) of dimensionality
Relevant Papers:
The use of multiple measurements in taxonomic problems by R. A. Fisher
On the generalized distance in statistics by P. C. Mahalanobis
Parzen's (1962) paper and its extension to R^d
Epanechnikov's paper on product kernels
OPTIMAL SMOOTHING IN KERNEL DISCRIMINANT ANALYSIS
ON ERROR-RATE ESTIMATION IN NONPARAMETRIC CLASSIFICATION
Choice of neighbor order in nearest-neighbor classification
Geometric representation of HDLSS data
Consistency of kNN (check p.2 onwards)
Assignments (old):
Numerical (Additional Information : 1 , 2 )
Assignments will be given, and discussed in the class (if necessary). However, no grading for assignments.
Grading and Exam policy:
Mark for each exam is stated in brackets.
MidSem - [30]
Numerical Assignment - [25][with presentation]
EndSem - [45]
Final Score = MidSem + Numerical Assignment + EndSem
Grading: A - F: based on score quantiles
Enjoy learning!