http://robowiki.net/w/index.php?action=history&feed=atom&title=Thread%3ATalk%3ABeepBoop%2FAwesome_enty%2Freply_%2817%29 2026-04-09T02:25:36Z Revision history for this page on the wiki MediaWiki 1.34.1 http://robowiki.net/w/index.php?title=Thread:Talk:BeepBoop/Awesome_enty/reply_(17)&diff=56678&oldid=prev 2021-06-22T18:38:57Z

<p>Reply to <a href="/wiki/Thread:Talk:BeepBoop/Awesome_enty/reply_(13)" title="Thread:Talk:BeepBoop/Awesome enty/reply (13)">Awesome enty</a></p> <p><b>New page</b></p><div>I tried a few (pretty simple) variants:<br /> * Multiplying the features by a weight matrix. One nice feature of this is that a diagonal matrix recovers standard feature weighting, so this model should be strictly better than per-feature weights.<br /> * A one-hidden-layer feedforward network.<br /> * Summing up the embeddings from the above two. <br /> I totally agree that allowing multiplicative feature interactions as you suggest should work better though!</div>

Kev