top | item 9011222

Text Understanding from Scratch Using Temporal Convolutional Networks

61 points| drewvolpe | 11 years ago |arxiv.org | reply

9 comments

order
[+] sgt101|11 years ago|reply
Astonishing? 3% better than bag of words after n days training on GPU's? I have misunderstood because I am not astonished.
[+] bradneuberg|11 years ago|reply
It's at the lower level character rather than word level which is unique. The convolutional net also doesn't need to be told what each word's role is, but rather learns that feature itself.
[+] eva1984|11 years ago|reply
What surprises me is that (BOW model + logistic regression) works just fine in most of the benchmarks(except for Amazon Review), interesting paper anyway. Could it be that because the vocabulary for BOW is limited to 5000, a lot of information is lost?
[+] ameasure|11 years ago|reply
Fascinating paper but the benchmarks seem incredibly weak. 5000 features for a bag of words model is nothing,these models normally have tens or hundreds of thousands of features.
[+] sushirain|11 years ago|reply
Open questions:

* Compare to RNNs with character level input.

* Compare to dedicated methods of sentiment analysis and topic categorization.

[+] petercooper|11 years ago|reply
This paper is pretty fascinating, thanks! Having trouble visualizing it but getting there..