Computer Information Services Department

TEACHING

Designed a course in Machine Learning/Statistics Interface (2007): Topics covered included: Clustering:  the interface between k-means, EM based clustering,  enhanced k-means clustering, Bayes Theorem:  Occam’s Razor and the reason for avoiding classical statistics. The advantages of Bayes theorem; Markov Chain Monte Carlo in Computational Analysis; Boosting in statistics and machine learning; The role of ‘distance’ and ‘density’ in formulating statistical models.  The special role of Kullback Leibler Divergence.; Sequential Markov Chain Monte Carlo:  Using Bayesian filters, particles to solve problems in inference; Robot Mapping and the alignment of maps

Desined a Seminar Course on 'Bayesian Statistics' (Spring, 1996) dealing with* (i) Basic Inference; (ii) Bayesian Testing and HPD regions; (iii) Gibbs sampling; (iv) The Schwartz and related criterions leading up to Bayes factors and predictive inference.

Shape Analysis and pattern recognition (see cvpr paper, tpami paper)

Time-Motion Orientation and Body Posture. (I am submitting 2 grants with Emily Keshner and Geoffrey Wright on this topic)

Sequential Markov Chain Monte Carlo Techniques (see IJCV paper, and JASA paper (in preparation)

Contour Grouping (see nips paper)

Multiresolution Markov Random Fields. (see nips paper)
·

Dirichlet Processes : Topics in Nonparametric Bayesian Inference.
· Markov Chain Monte Carlo Techniques : Using Markov Chain Enhanced        Simulation Techniques                                                                                
· Applications of Statistics to Marketing.
· Finance: Wavelets-based analysis of the influence of markets on one another.               

1. "A New Approach to the estimation of inter-variable correlation", Communications in Statistics 37: 2315-2330, 2008.

2. "Bayes and Empirical Bayes Procedures for Comparing Parameters", Journal of the American Statistical Association, 88, 687-693, (1993).

3. "New EM derived from Kullback-Leibler Divergence" , Conference on Knowledge and Data Discovery (KDD) 2006. (with Longin Jan Latecki and Rolf Lakaemper)

4 . "Bayes and Minimax Sample-Admissible Multivariate Selection Problems", Journal of the American Statistical Association, 86, 1022-1023, (1991).
5. Clustering by Distance Projection in the Feature Space with Applications to Video  Analysis (with Longin Latecki, Steve Horvath, Rajagopal) (in pdf) published in IEEE 2003 Conference on Data Mining.

6. "Correspondences between Parts of Shapes with Partical Filters" (with Rolf Lakaemper) published in CVPR (2008) [computer vision and pattern recognition].

7 . "Statistics of Natural Populations" (with J. Arnold and M. Sobel), Biometrics, 42, 45-65 (1986).

8 . "Cross-Validated Assessments of Test-Scoring Models", Communications in Statistics, 21 (11), 3037-3052, (1992).


9. "The Estimation of the Number of Balls Put in a Sequence of Urns", Communications in Statistics, 18 (4), 1297-1307 (1989) .

10. Data Visualization by Distance Distortion Minimization (to be published in a special volume of Communications (2004) in statistics in honor of Milton Sobel).

• "Parametric Piecewise Linear Models with a Good Fit to the Data", accepted for publication in IEEE Transactions on Pattern and Machine Intelligence (TPAMI) 2008/9

• "Bayesian Algorithms for Building Correspondences between Shapes" , soon to be under review by JASA.

• Clustering by Distance Projection in the Feature Space with          Applications to Video  Analysis  (with Longin Latecki, Steve Horvath, Rajagopal) (in pdf) published in IEEE 2003 Conference on Data Mining.

• Effective Feature Selection for Characterization and Classification of Spatial Region Data (published in  the 2004 conference on Artificial Intelligence)

• "Multiscale Random Fields with Application to Contour Grouping" (with Longin Jan Latecki) accepted for publication in NIPS (Neural information processing systems) 2008/2009.