 |
ML Newsnew | past | best | rss | submit | about |
| Neural Networks as Explicit Word-Based Rules (arxiv.org) | |
| 1 point by godelmachine on Jul 16, 2019 | hide | past | pdf | discuss on HN | |
|
In plain words: Each filter in a network that judges whether a review is positive or negative becomes a plain word rule, built from the word patterns it responds to most. Those rules match the original network's accuracy, so its decisions can be read as word patterns. AbstractFilters of convolutional networks used in computer vision are often visualized as image patches that maximize the response of the filter. We use the same approach to interpret weight matrices in simple architectures for natural language processing tasks. We interpret a convolutional network for sentiment classification as word-based rules. Using the rule, we recover the performance of the original model. | |
| add comment on HN |