How can we correlate the neural activity in the human brain as it responds to typed words, with properties of these terms (like 'edible', 'fits in hand')? In short, we want to find latent variables, that jointly explain both the brain activity, as well as the behavioral responses. This is one of many settings of the Coupled Matrix-Tensor Factorization (CMTF) problem. Can we accelerate any CMTF solver, so that it runs within a few minutes instead of tens of hours to a day, while maintaining good accuracy? We introduce TURBO-SMT, a meta-method capable of doing exactly that: it boosts the performance of any CMTF algorithm, by up to 200x, along with an up to 65 fold increase in sparsity, with comparable accuracy to the baseline. We apply TURBO-SMT to BRAIN Q, a dataset consisting of a (nouns, brain voxels, human subjects) tensor and a (nouns, properties) matrix, with coupling along the nouns dimension. TURBO-SMT is able to find meaningful latent variables, as well as to predict brain activity with competitive accuracy.
|Original language||English (US)|
|Title of host publication||SIAM International Conference on Data Mining 2014, SDM 2014|
|Editors||Mohammed J. Zaki, Arindam Banerjee, Srinivasan Parthasarathy, Pang Ning-Tan, Zoran Obradovic, Chandrika Kamath|
|Publisher||Society for Industrial and Applied Mathematics Publications|
|Number of pages||9|
|State||Published - 2014|
|Event||14th SIAM International Conference on Data Mining, SDM 2014 - Philadelphia, United States|
Duration: Apr 24 2014 → Apr 26 2014
|Name||SIAM International Conference on Data Mining 2014, SDM 2014|
|Other||14th SIAM International Conference on Data Mining, SDM 2014|
|Period||4/24/14 → 4/26/14|
Bibliographical noteFunding Information:
Research was funded by grants NSF IIS-1247489, NSF IIS- 1247632, NSF CDI 0835797, NIH/NICHD 12165321, and DARPA FA87501320005. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the funding parties. The authors would also like to thank Leila Wehbe and Alona Fyshe for their initial help with the BrainQ data.
Copyright © SIAM.