New work out on arXiv! Reverse engineering recurrent networks for sentiment classification reveals line attractor dynamics (https://arxiv.org/abs/1906.10720 ), with fantastic co-authors @ItsNeuronal, @MattGolub_Neuro, @SuryaGanguli and @SussilloDavid. #tweetprint summary below!
(1/4)
-
-
Is this effect specific to sentiment-analysis?
-
So far we’ve only analyzed binary sentiment classification, which we thought was a good first task to analyze. We hope to analyze more tasks in future work!
- Još 2 druga odgovora
Novi razgovor -
-
-
Finally, we can understand how the network processes individual words (tokens) by looking at projections of the embedding vectors onto the principal eigenvectors of the system. Overall, we think these tools will help us demystify and understand how recurrent networks work! (4/4)pic.twitter.com/BwVqB2n8SK
Prikaži ovu nitHvala. Twitter će to iskoristiti za poboljšanje vaše vremenske crte. PoništiPoništi
-
Čini se da učitavanje traje već neko vrijeme.
Twitter je možda preopterećen ili ima kratkotrajnih poteškoća u radu. Pokušajte ponovno ili potražite dodatne informacije u odjeljku Status Twittera.


