Papers by Guillermo Cecchi
Nearly-Unsupervised Hashcode Representations for Biomedical Relation Extraction (D19-1)
Copied to clipboard
| Challenge: | Recent studies have shown that locality sensitive hashcodes are useful for biomedical relation extraction tasks. |
| Approach: | They propose to optimize locality sensitive hashcode representations in a nearly unsupervised manner . they use only data points, but not their class labels, for learning . |
| Outcome: | The proposed approach improves accuracy from training to test sets, and the data points are only used for learning . |