Fixed-Length Poisson MRF: Adding Dependencies to the Multinomial
Authors: David I. Inouye, Pradeep K. Ravikumar, Inderjit S. Dhillon
NeurIPS 2015 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We show the effectiveness of our LPMRF distribution over Multinomial models by evaluating the test set perplexity on a dataset of abstracts and Wikipedia. Qualitatively, we show that the positive dependencies discovered by LPMRF are interesting and intuitive. |
| Researcher Affiliation | Academia | David I. Inouye Pradeep Ravikumar Inderjit S. Dhillon Department of Computer Science University of Texas at Austin {dinouye,pradeepr,inderjit}@cs.utexas.edu |
| Pseudocode | No | The paper describes algorithms in text but does not provide structured pseudocode or algorithm blocks. |
| Open Source Code | Yes | Finally, we show that our algorithms are fast and have good scaling (code available online). |
| Open Datasets | Yes | We evaluated our novel LPMRF model using perplexity on a held-out test set of documents from a corpus composed of research paper abstracts3 denoted Classic3 and a collection of Wikipedia documents. (Footnote 3: http://ir.dcs.gla.ac.uk/resources/test_collections/) |
| Dataset Splits | No | We train all the models using a 90% training split of the documents and compute the held-out perplexity on the remaining 10% where perplexity is equal to exp( L(Xtest|θ1...k, Φ1...k)/Ntest), where L is the log likelihood and Ntest is the total number of words in the test set. The paper specifies a 90/10 train/test split but does not explicitly mention a separate validation split. |
| Hardware Specification | Yes | All timing experiments were conducted on the TACC Maverick system with Intel Xeon E5-2680 v2 Ivy Bridge CPUs (2.80 GHz), 20 CPUs per node, and 12.8 GB memory per CPU (https://www.tacc.utexas.edu/). |
| Software Dependencies | No | In C++, we implemented the three core algorithms... trivially parallelized using Open MP (http://openmp.org/). For LDA, we used... a MATLAB. The paper mentions software but does not specify version numbers for any of them. |
| Experiment Setup | Yes | For a single Multinomial or LPMRF, we set the smoothing parameter β to 10 4. We select the LPMRF models using all combinations of 20 log spaced λ between 1 and 10 3, and 5 linearly spaced weighting function constants c between 1 and 2... For LDA, we used 2000 iterations and optimized the hyperparameters α and β using the likelihood of a tuning set. |