CS 8001
Intelligent Systems Reading Group
Schedule

Date
Readings

2009, August 25
  1. J. Peters, S. Vijayakumar, and S. Schaal (2003). Reinforcement Learning for Humanoid Robotics. ICHR.
Hey, it's about, um, Reinforcement Learning for Humanoid Robotics.


2009, August 18
  1. J. Zico Kolter and Andrew Y. Ng (2009). Near-Bayesian Exploration in Polynomial Time. ICML.
A regularization framework for LSTD. Mmmmm.


2009, August 17
We return after a year-long break
The Theme is Reinforcement Learning




2008, March 26
  1. Beygelzimer, Kakade, Langford (2006). Cover Trees for Nearest Neighbor. ICML.
...back to ICML
The interested reader may also look at: http://hunch.net/~jl/projects/cover_tree/cover_tree.html


2008, March 5
  1. Moghaddam, Weiss, Avidan (2006). Spectral Bounds for Sparse PCA: Exact and Greedy Bounds. NIPS
...back to NIPS.


2008, February 27
  1. Zhou, Hastie and Tibshirani (2006*). Sparse Principal Component Analysis. JCGS 15(2).
Continuing our series of papers that look longer than they are.

*Year of publication... actually written in 2004, apparently.



2008, February 20
  1. Tibshirani (2003*). A Simple Explanation of the Lasso and Least Angle Regression. Stolen from: http://www-stat.stanford.edu/~tibs/lasso/simple.html
  2. Zou and Hastie (2005). Regularization and Variable Selection Via the Elastic Net. J.R. Statist Soc. B.
One very short explanation of Lasso and then some elastic nets.

*Well, according to the way back machine, at any rate.



2008, February 13
  1. Tibshirani (1996). Regression Shrinkage and Selection via the Lasso. Journal of the Royal Statistical Society
It's 23 pages, but they're small pages.


2008, February 6
  1. Bishop (2006). Chapter 6: Kernel Methods. Pattern Recognition and Machine Learning


2008, January 30
  1. Weston, Mukherjee, Chapelle, Pontil, Poggio, Vapnik (2008). Feature Selection for SVMs. NIPS.


2008, January 23
  1. Siddiqi, Boots, Gordon (2008). A Constraint Generation Approach to Learning Stable Linear Dynamical Systems. NIPS.


2008, January 16
  1. Gashler, Ventura, and Martinez (2008). Iterative Non-linear Dimensionality Reduction by Manifold Sculpting. NIPS.


2008, January 9
  1. Cayton and Dasgupta (2008). A Learning Framework for Nearest Neighbor Search. NIPS.
Our first paper of the new term. This begins our series on new and popular algorithms / frameworks explored at NIPS 2007.




2007, October 17
  1. Minnen, Isbell, Essa and Starner (2007). Discovering Multivariate Motifs using Subsequence Density Estimation and Greedy Mixture Learning. AAAI.
More on activity discovery. We may spend some time on [2] from last week as well.


2007, October 10
  1. Keogh, Lin and Truppel (2003). Clustering of Time Series Subsequences is Meaningless: Implications for Previous and Future Research. ICDM.
  2. Chiu, Keogh and Lonardi (2003). Probabilistic Discovery of Time Series Motifs. SIGKDD.
Time to switch topics: activity discovery and recognition. These are good starter papers (we read them about three years ago) that get us started into one particular line of research.


2007, October 3
  1. Marthi, Russell, Latham, Guestrin (2005). Concurrent Hierarchical Reinforcement Learning. IJCAI.
Continuing this thread (as it were)....


2007, September 26
  1. Bhat, Isbell, Mateas (2006). On the Difficulty of Modular Reinforcement Learning for Real-World Partial Programming. AAAI.
Short followup on last week's paper. Bring [1] and [2] from last week as well.


2007, September 19
  1. S. Russell and A. Zimdars (2003). Q-Decomposition for Reinforcement Learning Agents. ICML-03. 2003.
  2. N. Sprague and D. Ballard (2003). Multiple-Goal Reinforcement Learning with Modular Sarsa(0). ICML.
Oldies but goodies, moving in the multi-agent RL problems space. [1] is our paper, but [2] give s a very nice overview and is much shorter.


2007, September 12
  1. Zhang, Aberdeen, Vishwanathan (2007). Conditional Random Fields for Multi-Agent Reinforcement Learning. ICML.

Continuing the theme of papers from this year's ICML.



2006, September 5
no ISRG


2007, August 29
  1. Salakhutdinov, Mnih, and Hinton (2007). Restricted Boltzmann Machines for Collaborative Filtering. ICML.

Continuing the theme of papers from this year's ICML.



2007, August 22
  1. Zhang, Xue, Sun, Guo, and Lu (2007). Optimal Dimensionality of Metric Space for Classification. ICML.

It is the beginning of a new term! We begin with papers from this year's ICML.





2007, April 11
  1. Choudhry and Basu (2004). Modeling Conversational Dynamics as a Mixed-Memory Markov Process. NIPS.

You know what we need? Machine learning papers applied to human activities.



2007, April 4
  1. Raj, Smaragdis, Shashanka (2006). Latent Dirichlet Decomposition for Single Channel Speaker Spearation. ICASSP.

You know what we need? Latent Dirichlet decompositions



2007, March 28
  1. Wang, Pentney, Papescu, Choudhury, and Philipose (2007). Common Sense Based Joint Training. IJCAI.

You know what we need? Machine learning papers applied to human activities.



2007, March 21
Spring Break


2007, March 14
  1. Shoham, Powers, and Grenager (2003).Multi-Agent Reinforcement Learning: A Critical Survey. Stanford Technical Report.

We're back!



2007, February 14
  1. Thomaz and Breazeal (2006).Reinforcement Learning with Human Teachers: Evidence of feedback and guidance with implications for learning performance. AAAI.
  2. Thomaz, Hoffman and Breazeal (2006).Reinforcement Learning with Human Teachers: Understanding how people want to teach robots.. ROMAN.

Machine Learning. Humans. Robots. Sheep. Wolves.



2007, Feburary 7
too many paper deadlines this week


2007, January 31
  1. Magerko and Laird (2003).Building an Interactive Drama Architecture. TIDSE.

Hm.



2007, January 24
  1. Ho and Pepyne (2001). Simple Explanation of the No Free Lunch Theorem of Optimization. IEEE D&C.
  2. Wolpert and Macready (1997). No Free Lunch Theorems for Optimization. IEEE Transactions.

We've decided that [2] is worth looking at.



2007, January 17
  1. Ho and Pepyne (2001). Simple Explanation of the No Free Lunch Theorem of Optimization. IEEE D&C.
  2. Wolpert and Macready (1997). No Free Lunch Theorems for Optimization. IEEE Transactions.

Well, you wanted it and we've got it: No Free Lunch. [1] is a shorter explanation of [2]. We're reading [1], but [2] may have be worth looking over.



2007, January 10
  1. Kearns, Mansour, Ng, Ron (1995). An Experimental and Theoretical Comparison of Model Selection Methods. COLT.
  2. Kearns, Mansour, Ng, Ron (1997). An Experimental and Theoretical Comparison of Model Selection Methods. Machine Learning 27.

A nice way to start off the term. We are reading [1]; however, the extremely interested reader may want to read [2], the longer (now with over 33% more notation!) journal article.




2006, November 29
  1. James, Singh and Littman (2004). Planning with Predictive State Representations. ICMLA.

So... can we plan with these things?



2006, November 15
  1. James and Singh (2004). Learning and Discovery of Predictive State Representations in Dynamical Systems with Reset. ICML.

So... can we learn the things?



2006, November 8
  1. Singh, James and Rudary (2004). Predictive State Representations: A New Theory for Modeling Dynamical Systems . UAI.

So we begin a few weeks on predictive state representations. Plus: My voice is back!



2006, November 1
  1. Balkcom and Mason (2004). Introducing robotic origami folding. ICRA.

This should get us ready for this week's RIM.



2006, October 18
  1. D. Roberts, M. Nelson, C. Isbell, M. Mateas, M. Littman (2006). Targeting Specific Distributions of Trajectories in MDPs. AAMAS.



2006, October 11
  1. White, Holloway (2006). Resolvability for Imprecise Multi-attribute Alternative Selection.



2006, October 4
no ISRG


2006, September 27
  1. Chajewska, Koller, Parr (2000). Making Rational Decisions Using Adaptive Utility Elicitation. AAAI.



2006, September 20
  1. Isbell, Kearns, Singh, Shelton, Stone, and Kormann (2006). Cobot in LambdaMOO: An Adaptive Social Statistics Agent. Autonmous Agents and Multi-Agent Systems.



2006, September 13
  1. Aha, Molineaux, and Ponsen (2005). Learning to Win: Case-based plan selection in a real-time strategy game. CCBR.

Finish the Aha paper.



2006, September 6
  1. Mateas and Stern (2003). Facade: An Experiment in Building a Fully-Realized Interactive Drama. Game Developer's Conference.
  2. Aha, Molineaux, and Ponsen (2005). Learning to Win: Case-based plan selection in a real-time strategy game. CCBR.

We will be reading two papers in the space of interesting games with AI. Read the first paper and the first two sections of the second. Next week we will finish the second paper and look at another AI/ML approach to building players of games.



2006, August 30
  1. Blekas, Fotiadis, and Likas (2003). Greedy mixture learning for multiple motif discovery in biological sequences. Bioinformatics.


2006, August 23
plans for the term




2006, April 17
qualifer exams today, so no ISRG


2006, April 10
  1. E. Altendorf, A. Restificar, T. Dietterich (2005). Learning from Sparse Data by Exploiting Monotonicity Constraints. UAI.
ISRG will be held in TSRB 133 Monday at 1pm instead of in the usual place. Angelo Restificar will be giving a talk. The paper is background reading.


2006, April 3
  1. D. Jensen and J. Neville (2002). Linkage and Autocorrelation Cause Feature Selection Bias in Relational Learning. ICML.
  2. J. Neville, D. Jensen, L. Friedland and M. Hay (2003). Learning Relational Probability Trees. SIGKDD.


2006, March 27
  1. J. Neville and D. Jensen (2004). Dependency Networks for Relational Data. ICDM.


2006, March 20
Spring Break


2006, March 6
  1. C. Yu, D. Ballard and R. Aslin (2003). The Role of Embodied Intention in Early Lexical Acquisition. Gog Sci 2003.


2006, February 27
  1. Cohen, Oates, Beal, Adams (2002). Contentful Mental States for Robot Baby. AAAI 2002.
  2. Sebastiani, Ramoni and Cohen (1999). Bayesian Clustering of Sensory Inputs by Dyanmics. UMASS Tech Report.


2006, February 20
we skip a week while everyone works on AAAI papers


2006, February 13
  1. Nelson, Roberts, Isbell, and Mateas (2006). Reinforcement Learning for Declarative Optimization-Based Drama Management. AAMAS 2006.
  2. Nelson and Mateas (2005). Search-based drama management in the interactive fiction Anchorhead. AIIDE-05.
Intelligent Entertainment. [1] is the main paper, [2] may help you with some of the background.


2006, February 6
  1. W. Maass, R. A. Legenstein, and N. Bertschinger (2005). Methods for estimating the computational power and generalization capability of neural microcircuits. NIPS.
back to neural stuff


2006, January 30
  1. R. O'Donnell (2005). Learning monotone functions from random examples in polynomial time.
A learning theory paper! Followed by a talk at 2pm in the Learning Seminar at MiRC 102.


2006, January 23
  1. R. Rao and T. Sejnowski (2001). Spike-Timing-Dependent Hebbian Plasticity as Temporal Difference Learning. Neural Computation.
  2. A. Shon, R. Rao and T. Sejnowski (2004). Motion Detection and Prediction through Spike-Timing Dependent Plasticity.
Our first in a series of neural-inspired learning. The first paper demonstrates that a model of neural computation simulates temporal difference learning while the second is an application of those sorts of ideas to a the problem of motion detection and prediction.




2005, December 5
  1. A. von Hessling, A. Goel (2005). Abstracting Reusable Cases from Reinforcement Learning. ICCBR05.
Remember back in late October when we were to have two talks on structural learning but we ended up with only one? Today is the other one. The paper above is background reading.


2005, November 28
No paper today; however, there is a talk that should complement our recent discussions at 2pm in TSRB 132:
The Holy Grail of MMOG Design: Supporting Players to Perform


2005, November 21
  1. S. Russell and A. Zimdars (2003). Q-Decomposition for Reinforcement Learning Agents. ICML-03. 2003.
  2. N. Sprague and D. Ballard (2003). Multiple-Goal Reinforcement Learning with Modular Sarsa(0). ICML.
[1] is our paper, but [2] give s a very nice overview and is much shorter.


2005, November 14
  1. Dietterich, T (2000). An Overview of MAXQ Hierarchical Reinforcement. SARA.
MAXQ!


2005, November 7
  1. D. Andre and S. Russell (2002). State Abstraction for Programmable Reinforcement Learning Agents.. AAAI.
We continue looking at adaptive programming languages and the issues that arise.


2005, October 31
  1. P. Norvig and D. Cohn. Adaptive Software.
  2. D. Andre and S. Russell (2001). Programmable Reinforcement Learning Agents. NIPS.
We begin our series on adaptive or partial programming. The first paper provides an overview and background and the second is the first a few papers from this group that deals with issues that arise when you want to do this kind of learning inside of programming.


2005, October 24
  1. A. von Hessling, A. Goel (2005). Abstracting Reusable Cases from Reinforcement Learning. ICCBR05.
  2. J. Jones, A. Goel (2005). Knowledge Organization and Structural Credit Assignment. Workshop at IJCAI.
In a change of pace, we'll have two presentations on research related to some of the structural learning papers we've been reading over the last few weeks. The papers above are background reading.


2005, October 17
Fall Break


2005, October 10
  1. K. Driessens, J. Ramon, H. Blockeel (2001). Speeding up Relational Reinforcement Learning Through the User of an Incremental First Order Decision Tree Learner. EMCL.


2005, October 3
Free Day


2005, September 26
  1. C. Guestrin, D. Koller, C. Gerahart, N. Kanodia (2003). Generalizing Plans to New Environments in Relational MDPs . IJCAI.


2005, September 19
  1. K. Nigam, J. Lafferty, A. McCallum (1999). Using Maximum Entropy for Text Classification. Workshop at IJCAI.
  2. P. Penfield (2003). Principle of Maximum Entropy: Simple Form. Class Notes MIT 6.050.
  3. P. Penfield (2003). Principle of Maximum Entropy. Class Notes MIT 6.050.
Hm. Yes, well. Let's try that again. [1] is the paper, [2] & [3] are background that you might find helpful.


2005, September 12
  1. T. Jaakkola, M. Meila, & T. Jabara (2000). Maximum Entropy Discrimination. NIPS 1999.
Because I want talk about Jaakkola.


2005, September 5
Labor Day


2005, August 29
  1. C. Bishop and M. Tipping (2000). Variational Relevance Vector Machine. UAI 2000.
It's next time.


2005, August 22
  1. M. Tipping (2000). The Relevance Vector Machine. NIPS 1999.
A probabilistic / bayesian treatment of SVMs. More next time.


2005, April 25
  1. E. Kiciman and A. Fox (to appear). Detecting Application-Level Failures in Component-based Internet Services. IEEE Transactions on Neural Networks: Special Issue on Adaptive Systems, Spring 2005.


2005, April 18
  1. C. Park and H. Park (to appear). Nonlinear discriminant analysis using kernel functions and the generalized singular value decomposition. SIAM Journal on Matrix Analysis and Applications.


2005, April 11
  1. A. Apostolico and G. Bejerano (2000). Optimal Amnesic Probabilistic Automata Journal of Computational Biology, 7(3/4).


2005, April 4
Quals are today


2005, March 28
Alex Gray will be visiting us in class


2005, March 21
Spring Break


2005, March 14
  1. A. Gray and A Moore (2004). N-Body Problems in Statistical Learning NIPS 2004.


2005, March 7
  1. A Josang and Pope (2005). Semantic Constraints for Trust Transitivity. APCCM 2005.
Uh-oh. An application. Must mean it's ending.


2005, February 28
  1. A Josang (2001). A Logic for Uncertain Probabilities. International Journal of Uncertainty, Fuzziness and Knowledge-Based Systems 9(3).
It never ends....


2005, February 21
  1. P. Wang (2001). Confidence as Higher-Order Uncertainty 2nd International Symposium on Imprecise Probabilitie and Their Applictions.
The next salvo.


2005, February 14
  1. P. Wang (2004). The Limitation of Bayesianism Artificial Intelligence 158(1).
Holy war!


2005, February 7
  1. Andrew Ng and H Jin Kim (2005). Stable Adaptive Control with Online Learning NIPS 04.
So here's an idea for adapting to changing circumstances without doing something too dumb along the way.


2005, January 31
  1. M. Kaess, R. Zboinski, and F. Dellaert (2004). Multiview Reconstruction of Piecewise Smooth Subdivision Curves with a Variable Number of Control Points ECCV 04.
  2. Green, with discussion by Godsill and Heikkinen (2003). Trans-dimensional Markov chain Monte Carlo chapter for the book on Highly Structured Stochastic Systems.
We are reading [1]. Here's your chance to mock Frank in a public forum provided you can understand the math. [2] is background you might find useful to refer to for that purpose, but Frank does warn that it's quite "advanced". So far, I'd say he's up one point on the mocking.


2005, January 24
  1. Rish (2001). An empirical study of the naive Bayes classifier IJCAI-01 workshop on "Empirical Methods in AI". Also appeared as IBM Technical Report RC22230.
  2. Pedro Domingos and Michael Pazzani (1997). On the Optimality of the Simple Bayesian Classifier under Zero-One Loss Machine Learning, 29, 103-130.
Both [1] and [2] try to characterize the conditions under which "simple" Bayesian classifiers are optimal, other than the obvious one. Come prepared to discuss their findings and with an opinion on their analyses and techniques.


2004, January 17
MLK Holiday


2005, January 10
  1. N. Hill, T. Lal, K Bierig, N Birbaumer and B. Scholkopf (2005). An Auditory Paradigm for Brain-Computer Interfaces. NIPS 17.
Our first meeting is 1pm this Monday at TSRB 223, where we will be all term




2004, November 29
And so the term ends. See you in January.
Note: there will be an actual reading for the first week of the term, so check back here the week before.


2004, November 22
  1. P. Atkar, D. Conner, A. Greenfield, H. Choset, and A. Rizzi (2004). Uniform Coverage of Simple Surfaces Embedded in R3 for Auto-Body Painting. Sixth Workshop on the Algorithmic Foundations of Robotics, Utrecht/Zeist, The Netherlands, 2004.
You know what we don't talk about enough? Problems in Robotics.


2004, November 15
  1. J Fogarty, S. Hudson, and J. Lai (2004). Examining the Robustness of Sensor-Based Statistical Models of Human Interruptibility. Proceedings of the ACM Conference on Human Factors in Computing Systems, pp. 207-2214.
  2. S Hudson, et al (2003). Predicting Human Interruptibility with Sensors: A Wizard of Oz Feasibility Study. Proceedings of the ACM Conference on Human Factors in Computing Systems, pp. 257-264.
Here we examine a different subfield of CS's contribution to a related field of activity discovery by looking at an HCI/ML approach to modeling interruptibility. [1] is about an actual system while [2] is a "Wizard of Oz" pilot study that inspired it.


2004, November 8
  1. M. Holmes and C. Isbell (2005). Schema Learning: Experience-Based Construction of Predictive Action Models . NIPS 17.
We return to something more like activity discovery with a touch of PSRs.


2004, November 1
  1. R. Sutton and B. Tanner (2005). TD Networks. NIPS 2004.
This is the almost-final version of a paper to appear in NIPS 2005. The final version will appear Monday, probably right after we finish talking about this version. In any case, it's a generalization of TD learning and has connections to PSRs.


2004, October 24
Skipping a week


2004, October 17
Fall Break


2004, October 11
  1. V. Pavlovic, J. M. Rehg, and J. MacCormick (2001). Learning Switching Linear Models of Human Motion. NIPS 2000.
More Quals Reading List.


2004, October 4
  1. F. Dellaert, S. Seitz, C. Thorpe, and S. Thrun (2001). Feature Correspondence: A Markov Chain Monte Carlo Approach. NIPS 2001.
  2. MacKay (1998). Introduction to Monte Carlo Methods. Learning in Graphial Models, pp 175-204.
Right. The Quals Reading List for Intelligent Systems. Here's one of the papers using MCMC for feature correspondence. I'll let Frank know if any of you misunderstand it, so be on your toes. The second paper is for background. Many former quals takers have found it to be extremely helpful.


2004, September 27
  1. P. Stone, M. Littman, S. Singh and M. Kearns (2000). ATTac-2000: An Adaptive Autonomous Bidding Agent. JAIR 15:189-206.
  2. N. Montfort (2003). Arby: An Agent to Assist in Bargaining and Mediation. Poster/Presentation at Penn Engineering Graduate Research Symposium.
  3. N. Montfort (2003). Decision and Learning in Computational Behavorial Game Theory. Poster at ICML-2003.
We continue our game theory discussion this week by looking at two applied systems that use game theory principles and/or live in a domain that is typically modeled using game theory. [2] and [3] are one page posters, and so should be easy reads; however, they reveal some very serious opportunities and issues in practical applications of game theory.


2004, September 20
  1. C. Shelton (2000). Balancing Multiple Sources of Reward in Reinforcement Learning. NIPS 2000.
  2. M. Kearns (2002). Tutorial Slides on Computational Game Theory. NIPS 2002.
So, we take a break from activity discovery and begin a discussion on game theory this week and the next. [1] is a paper on balancing multiple sources of reward and is the paper we will be discussing. [2] is a set of slides given at a NIPS tutorial on computational game theory by Michael "Sauve and Smart" Kearns. You will find it quite useful as background, and almost certainly want to read it first.


2004, September 13
  1. S. Basu, M Bilenko, R Mooney (2004). A Probabilistic Framework for Semi-Supervised Clustering. KDD 2004.
Semi-supervised clustering. I'll let you work out the connections to activity and pattern discovery. Dense. I think it won a best paper award.


2004, August 30
  1. J. B. Tenenbaum, V. De Silva, J. C. Langford (2000). A global geometric framework for nonlinear dimensionality reduction. Science 290 (5500).
  2. Odest C. Jenkins and Maja J Mataric (2004).A Spatio-temporal Extension to Isomap Nonlinear Dimension Reduction. Proceedings, International Conference on Machine Learning (ICML-2004).
Papers on Isomap. So, we will be doing some non-linear dimensionality reduction with recent additions to handle time. Some of you will have read the first paper, but probably not the second.


2004, August 23
Class is now in TSRB 223
  1. Ajo Fod, Maja J Mataric, and Odest Chadwicke Jenkins. (2002) Automated Derivation of Primitives for Movement Classification, Autonomous Robots,12(1), 39-54.
Just one paper today, but it's seventeen pages long. We'll be returning to Jenkins and Mataric again.


2004, August 16
Boggs 339? What's up with that?
Initial planning decisions




2004, April 19
No class
Go finish your projects


2004, April 12
  1. Patterson, Liao, Fox and Kautz (2003). Inferring High-Level Behavior from Low-Level Sensors. Ubicomp.


2004, April 5
  1. Keogh, Lin and Truppel (2003). Clustering of Time Series Subsequences is Meaningless: Implications for Previous and Future Research. ICDM.


2004, March 29
  1. Oates (2002). PERUSE: An Unsupervised Algorithm for Finding Recurring Patterns in Time. ICDM.
  2. Chiu, Keogh and Lonardi (2003). Probabilistic Discovery of Time Series Motifs. SIGKDD.


2004, March 22
  1. Gentner. Exhuming Similarity.
  2. Heit. What is the probability of the Bayesian model, given the data?.
  3. Boroditsky and Ramscar. "First, we assume a spherical cow...".
  4. Love. Three deadly sins of category learning modelers.
  5. Tenenbaum and Griffiths (2001). Generalization, similarity and Bayesian inference. Behavorial and Brain Sciences 24, 629-640.
We are discussing [2] from last week as well as several short responses to it (each is something like a page). [5] is a longer version of [2] from last week that might be useful for reference.


2004, March 15
  1. J. Tenenbaum (1999). Bayesian Modeling of Human Concept Learning. NIPS 1999.
  2. J. Tenenbaum (2000). Rules and Similarity in Concept Learning. NIPS 2000.


2004, March 8
Spring Break


2004, March 1
see previous week
As mentioned at our meeting of February 23, we will continue our discussion of both [1] and [2], focusing on technical details.


2004, February 23
  1. Y. Ivanov and B. Blumberg (2002). Solving Weak Transduction with Expectation Maximization. Robotics and Autonomous Systems 972:1-15.
  2. R. Burke and B. Blumberg (2002). Using an Ethologically-Inspired Model to Learn Apparent Temporal Causality for Planning in Synthetic Creatures. Proceedings of AAMAS.
We are now moving into notions of learning temporal clusters and such. Read [1] thoroughly. Then read [2] thoroughly. Have deep insightful comments, because I really want to understand this stuff and if you don't help me, I swear I will destroy you all. We may take this into next week as well (this is one of those interesting traps: if you read [1] and keep an interesting discussion going then there is no reason to read [2] because we'll discuss it next week... but if you don't read [1] throughly then you'll also have to read [2] because we won't sustain a discussion for an hour about [1]... personally, I suggest reading [1] thoroughly).


2004, February 16
  1. N. Tishby, F. Pereira, and W. Bialek (1999). The Information Bottleneck Method. The 37th annual Allerton Conference on Communication, Control and Computing.
  2. N. Slonim and N. Tishby (2000). Document Clustering using Word Clusters via the Information Bottleneck Method. Proceedings of SIGIR.
[1] is from the reading list and [2] is an application of the technique. Might be worth comparing to [2004-02-02[2]].


2004, February 9
  1. B. Draper, K. Baek, M. Bartlett, J. Beveridge (2003). Recognizing Faces with PCA and ICA. Computer Vision and Image Understanding (91)115-137.
We are continuing our discussion about ICA, PCA and other As from last week. Given the chaos of the big move on 2/2 you should expect to also discuss papers [1] and [2] from last week.


2004, February 2
  1. A. Bell and T. Sejnowski (1996). Edges are the 'Independent Components' of Natural Scenes. NIPS 1996. An extended version from Vision Research 37(23) 3327-3338 is also available.
  2. C. Isbell and P. Viola (1998). Restructuring Sparse High Dimensional Data for Effective Retrieval. NIPS 1998.
  3. A. Bell and T. Sejnowski (1995). An information maximisation approach to blind separation and blind deconvolution. Neural Computation, 7, 6, 1129-1159.
[2] and [3] are from the reading list. [3] is reference zero for ICA, while [1] and [2] are applications of ICA. Read [2] and [3] carefully, and use [3] for background for this discussion if you have further questions about the ICA algorithm.


2004, January 26
  1. J. Kleinberg (2002). An Impossibility Result for Clustering. Proceedings of NIPS, 2002.
  2. P. Berkhin (2002). Survey Of Clustering Data Mining Techniques .
[1] is from the reading list. [2] provides some background. For Monday, pay attention to the first three sections, and think of the rest as reference material.


2004, January 26
MLK Holiday


2004, January 12
  1. M. Littman, R. Sutton, and S. Singh (2001). Predictive Representations of State. Proceedings of NIPS, 2001.
  2. S. Singh, M. Littman, N. Jong, D. Pardoe and P. Stone (2003). Learning Predictive State Representations. Proceedings of ICML, 2003.
  3. (optional) R. Rivest and R. Schapire (1994). Diversity-Based Inference of Finite Automata. Journal of the ACM, 1994.
[3] is background. It's optional, but worth referring to as you read [1] and [2]. [1] should be read before [2] but [3] can be read (skimmed, whatever) before, after, or during.