My Writings

This is where I keep my web-published writings on research and related topics. The content varies from rants, to notes to myself, to explanations of well-known techniques, to original research that will appear in other papers.

To cite any of these works, please use a citation like this:

@misc{RennieF04
,author = "Jason D. M. Rennie"
,title = "Derivation of the F-measure"
,year = 2004
,month = "February"
,howpublished = "http://people.csail.mit.edu/jrennie/writing"
}
Modeling Protein Interaction Experiments with a Trace Norm Prior.
Jason D. M. Rennie. December 20, 2006.
A Comparison of McCullagh's Proportional Odds Model to Modern Ordinal Regression Algorithms.
Jason D. M. Rennie. October 24, 2006.
Gradient Calculations for Factor Analysis and Selected Variations.
Jason D. M. Rennie. September 12, 2006.
Gradient Calculations for the Mean, Covariance Matrix Parameterization of the Multivariate Normal.
Jason D. M. Rennie. August 18, 2006.
Factor Analysis with a Trace Norm Prior.
Jason D. M. Rennie. June 20, 2006.
The Generalized Trace Norm.
Jason D. M. Rennie. April 19, 2006.
Text Modeling with the Trace Norm.
Jason D. M. Rennie. April 14, 2006.
Trace Norm is Less than the Sum of Column (or Row) Lengths.
Jason D. M. Rennie. March 11, 2006.
The Relation Between the Spectral and Trace Norms.
Jason D. M. Rennie. March 11, 2006.
Jacobian of the Singular Value Decomposition with Application to the Trace Norm Distribution.
Jason D. M. Rennie. February 27, 2006.
The Trace Norm Inequality and its Consequences.
Jason D. M. Rennie. February 27, 2006.
Model Selection is a Linear Function of λ.
Jason D. M. Rennie. February 27, 2006.
Learning Structure with the Trace Norm Distribution.
Jason D. M. Rennie. February 26, 2006.
Objective and Derivatives for MMMF using the Natural Parameter Multinomial.
Jason D. M. Rennie. February 22, 2006.
A Simple Exercise on Matrix Derivatives.
Jason D. M. Rennie. February 18, 2006.
Scaling the Trace Norm Distribution Partition Function.
Jason D. M. Rennie. February 4, 2006.
Estimating the Log-Partition Function of the Trace Norm Distribution.
Jason D. M. Rennie. February 2, 2006.
A Convexity Proof for the Partition Function of the Trace Norm Distribution.
Jason D. M. Rennie. February 1, 2006.
Computing the Trace Norm Distribution via Sampling.
Jason D. M. Rennie. January 30, 2006.
Approximating the Trace Norm Distribution Partition Function.
Jason D. M. Rennie. January 28, 2006.
Volume of the Stiefel Manifold.
Jason D. M. Rennie. January 24, 2006.
Toward Normalization of the Trace Norm Distribution.
Jason D. M. Rennie. January 21, 2006.
Whence the Determinant?.
Jason D. M. Rennie. January 10, 2006.
Volume of the n-sphere.
Jason D. M. Rennie. November 22, 2005.
Some Hypersphere Volumes.
Jason D. M. Rennie. November 22, 2005.
Trigonometric Integrals.
Jason D. M. Rennie. November 16, 2005.
The Log-Norm Distribution.
Jason D. M. Rennie. November 8, 2005.
The Gaussian-like Normalization Constant.
Jason D. M. Rennie. November 6, 2005.
On Trace Norm Regularization for Document Modeling.
Jason D. M. Rennie. November 1, 2005.
Another Hierarchical Topic Model.
Jason D. M. Rennie. October 27, 2005.
A Hierarchical Topic Model.
Jason D. M. Rennie. September 28, 2005.
Topics.
Jason D. M. Rennie. September 21, 2005.
Equivalent Ways of Expressing the Trace Norm of a Matrix.
Jason D. M. Rennie. September 8, 2005.
Mixtures of Multinomials.
Jason D. M. Rennie. September 1, 2005.
Matrix Factorizations are Unique only up to an Invertible Transformation.
Jason D. M. Rennie. August 31, 2005.
Relating the Trace and Frobenius Matrix Norms.
Jason D. M. Rennie. August 31, 2005.
Some Linear Algebra Notes.
Jason D. M. Rennie. August 26, 2005.
A Max-Margin Best-Previous-Match Coreference Resolution Algorithm.
Jason D. M. Rennie. July 25, 2005.
The Log-Log Term Frequency Distribution.
Jason D. M. Rennie. July 12, 2005.
The Unigram Term Frequency Distribution.
Jason D. M. Rennie. June 18, 2005.
Calculating Significance for a Single Test-Train Split.
Jason D. M. Rennie. June 8, 2005.
Calculation of the Hessian for the Log-Log Frequency Rate Distribution.
Jason D. M. Rennie. June 2, 2005.
A Role-Reversal in the Log-Log Model.
Jason D. M. Rennie. May 15, 2005.
A Class of Convex Functions.
Jason D. M. Rennie. May 13, 2005.
Using a Log-Log Distribution to Model Term Frequency Rates.
Jason D. M. Rennie. May 8, 2005.
Learning a Log-Log Term Frequency Model.
Jason D. M. Rennie. May 2, 2005.
A Better Model for Term Frequencies.
Jason D. M. Rennie. April 7, 2005.
Sparse Large Margin Matrix Factorization.
Jason D. M. Rennie. March 11, 2005.
Optimization of a Locally Convex Objective on Convex Regions.
Jason D. M. Rennie. March 11, 2005.
A Hybrid Model for Co-reference Resolution.
Jason D. M. Rennie. March 11, 2005.
A Bayesian 2-Mixture Model.
Jason D. M. Rennie. March 11, 2005.
Optimizing the MMMF Primal.
Jason D. M. Rennie. March 17, 2005.
Ordinal Smooth Hinge Classification.
Jason D. M. Rennie. February 22, 2005.
Smooth Hinge Classification.
Jason D. M. Rennie. February 22, 2005.
Modified Regularized Least Squares Classification.
Jason D. M. Rennie. February 14, 2005.
Ordinal Logistic Regression.
Jason D. M. Rennie. February 16, 2005.
Maximum-Margin Logistic Regression.
Jason D. M. Rennie. February 10, 2005.
The MMMF Objective: Primal and Dual.
Jason D. M. Rennie. February 4, 2005.
Ranking Sports Teams.
Jason D. M. Rennie. February 2, 2005.
The Conditional Random Field.
Jason D. M. Rennie. January 21, 2005.
How to Normalize a Kernel.
Jason D. M. Rennie. January 16, 2005.
Kernelized Softmax.
Jason D. M. Rennie. January 15, 2005.
Regularized Logistic Regression is Strictly Convex.
Jason D. M. Rennie. January 6, 2005.
Max-Antecedent is Convex Plus Concave.
Jason D. M. Rennie. January 4, 2005.
Learning Parameters for a Max-Antecedent Co-reference Resolution Model.
Jason D. M. Rennie. December 1, 2004.
A Max-Antecedent Model for Co-reference Resolution.
Jason D. M. Rennie. November 14, 2004.
Inferring a Noun Phrase Clustering for an Antecedent-based Co-reference Resolution Model.
Jason D. M. Rennie. November 12, 2004.
Learning Parameters for an Antecedent-based Co-reference Resolution Model.
Jason D. M. Rennie. November 11, 2004.
Why Sums are Bad.
Jason D. M. Rennie. November 4, 2004.
An Extraordinarily Brief Description of a Hybrid Model for Co-reference Resolution.
Jason D. M. Rennie. November 4, 2004.
A Rating Formulation of the SVM.
Jason D. M. Rennie. October 21, 2004.
You can find a better explanation in the Shashua/Levin (2003) paper.
The Equivalence of the Model of McCallum and Wellner (2003) to Minimum Cut and Maximum Inter-Cluster Similarity.
Jason D. M. Rennie. October 18, 2004.
Learning how to cluster with application to coreference resolution.
Jason D. M. Rennie. August 13, 2004.
Interesting Tasks for Restaurant Discussion Bulletin Boards.
Jason D. M. Rennie. July 13, 2004.
Restaurant Review Problem Statement.
Jason D. M. Rennie. July 7, 2004.
Derivatives for a Two-component Binomial Mixture Model.
Jason D. M. Rennie. July 1, 2004.
A Short Tutorial on Using Expectation-Maximization with Mixture Models.
Jason D. M. Rennie. March 3, 2004.
Derivation of the F-measure.
Jason D. M. Rennie. February 19, 2004.
Fast Leave-one-out Cross-validation for Regularized Least Squares Classification.
Jason D. M. Rennie. January 15, 2004.
Using Part-of-Speech Information for Transfer in Text Classification.
Jason D. M. Rennie. December 17, 2003.
On the Value of Leave-One-Out Cross Validation Bounds.
Jason D. M. Rennie. December 15, 2003.
Kernelizing Linear Classifiers.
Jason D. M. Rennie. December 6, 2003.
The Use of Transfer in Natural Language Processing.
Jason D. M. Rennie. December 2, 2003.
Regularized Least Squares Classification with a Gaussian Regularization Prior.
Jason D. M. Rennie. November 23, 2003.
Solving Mx=y .
Jason D. M. Rennie. November 20, 2003.
Learning More With Less.
Jason D. M. Rennie. July 17, 2003.
Newton's Method.
Jason D. M. Rennie. May 25, 2003.
Encoding Model Parameters.
Jason D. M. Rennie. May 23, 2003.
Stochastic encoding and the ``bits-back'' argument.
Jason D. M. Rennie. May 16, 2003.
NOTE: Argument isn't clean but the result is correct---KL-divergence determines the encoding length.
On L2-norm regularization and the Gaussian prior.
Jason D. M. Rennie. May 8, 2003.
Logistic regression.
Jason D. M. Rennie. April 23, 2003.
Boosting with decision stumps and binary features.
Jason D. M. Rennie. April 10, 2003.
Bounded loss classification.
Jason D. M. Rennie. February 25, 2003.
This paper also address the problem of stability.
One-versus-all alters Naive Bayes.
Jason D. M. Rennie. February 18, 2003.


Last modified: Wed Dec 20 00:14:31 2006