Comp 250MLS - Machine Learning Seminar
Department of Computer Science
Tufts University

Announcement(s): Please note time and room change: the course now meets on Mondays 10:30-11:45 at Halligan H-127.

Syllabus:

Schedule and papers

Date	Paper	Discussion
9/11: 10:30-11:45	Nearest Neighbors in high dimensions ICML2009 paper #360	Roni
Monday 9/14, 10:30-11:45	Topic Models Review Paper	Roni
9/21, 10:30-11:45	Author topic model KDD2004 Paper and UAI04 paper	Kevin
Wed 9/30 [note changed day], 10:30-11:45	Citation Topic Models ICML07 Paper #257 Relational Topic Models AISTAT2009 Paper Link Topic Models KDD2008 Paper	Byron Saket Saeed
Mon 10/5	A Mixture Model and EM-Based Algorithm for Class Discovery (link via Tufts domain to get access)	Bernie
Tue 10/13 (Tufts Monday schedule), 10:30-11:45	Semi-supervised and constrained prediction problems	Bilal and Bernie
Monday 10/19, 10:30-11:45	Penalized Probabilistic Clustering	Dan
Monday 10/26, 10:30-11:45	An Introduction to Variational Methods for Graphical Models	Roni
Monday 11/2, 10:30-11:45	Pairwise, Instance-Level Class Constraints (link via Tufts domain to get access)	Bilal
Monday 11/9, 10:30-11:45	LDA	Umaa
Monday 11/16, 10:30-11:45	Unsupervised Rank Aggregation	Kevin
Monday 11/23, 10:30-11:45	Supervised Rank Aggregation Semi-Supervised Ensemble Ranking	Saeed and Kevin
Monday 11/30, 10:30-11:45	SVM Optimization: Inverse Dependence on Training Set Size ICML2008 paper #266	TBA
Monday 12/7, 10:30-11:45	Generalized Bradley-Terry Models and Multi-class Probability Estimates	Umaa
Monday 12/14, 10:30-11:45	TBA	TBA

Additional Potential topics and papers

Topic	Papers	Notes
Misc. Topics
	Unify supervised Unsupervised ICML2009 paper #578
	Evluate Clustering ICML2009 paper #10
	Time Series ICML2009 paper #503
	Hidden Attributes ICML2009 paper #272
Topic Models
	LDA Paper
	ICML2009 paper #356
	ICML2009 paper #379
	ICML2009 paper #162
	ICML2009 paper #390
	Fast sampling KDD2008 Paper
Rank Aggregation
	Unsupervised ICML2008 paper #343
SVM, perceptron and related algorithms
	SVM training time ICML2008 paper #266
	Pegasos: Primal Estimated sub-GrAdient SOlver for SVM ICML07 Paper #587
	on line SVM KDD2008 Paper
	on line SVM Paper
	SVM reject option NIPS 2008 Paper #939
	on line ICML2009 paper #380
	on line ICML2009 paper #472
	on line application ICML2009 paper #42
	MKL ICML2009 paper #149
	MKL ICML2008 paper #158
	MKL ICML2008 paper #165
	MKL non PSD ICML2009 paper #520
	non PSD ICML2009 paper #399
	non PSD ICML2008 paper #531
Active Learning
	ICML2009 paper #392
	ICML2009 paper #262
	ICML2009 paper #475
	ICML2008 paper #519
	ICML2008 paper #324
Structured output spaces
	SEARN
	ICML2009 paper #420
	ICML2009 paper #297
Prediction with Graphs
	entailment in NLP Paper
	ICML2009 paper #485
	ICML2009 paper #542
SRL
	Learning directed Paper
	undirected ICML2008 paper #530
Link Prediction
	The Link-Prediction Problem for Social Networks
Learning with Label Constraints
	A Mixture Model and EM-Based Algorithm for Class Discovery (link via Tufts domain to get access)
	Pairwise, Instance-Level Class Constraints (link via Tufts domain to get access)
	Penalized Probabilistic Clustering
	Probabilistic Semi-Supervised Clustering with Constraints