Turbo-SMT: Accelerating Coupled Sparse Matrix-Tensor Factorizations by 200
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...
Saved in:
| Published in: | Proceedings of the ... SIAM International Conference on Data Mining Vol. 2014; p. 118 |
|---|---|
| Main Authors: | , , , , , |
| Format: | Journal Article |
| Language: | English |
| Published: |
United States
2014
|
| ISSN: | 2167-0102 |
| Online Access: | Get more information |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | 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
(CMTF) problem. Can we accelerate
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
CMTF algorithm, by up to
×, along with an up to
increase in sparsity, with comparable accuracy to the baseline. We apply TURBO-SMT to BRAINQ, 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. |
|---|---|
| Bibliography: | ObjectType-Article-1 SourceType-Scholarly Journals-1 ObjectType-Feature-2 content type line 23 |
| ISSN: | 2167-0102 |
| DOI: | 10.1137/1.9781611973440.14 |