armcat

15 ms·

1. 241.▲
   by armcat 5y ago

   You are welcome :-). With text we have fairly discrete and granular units, i.e. "words", and we see sentences as compositions of words. For images however, it's not so simple. We present the entire image (array of pixels time
2. 242.▲
   by armcat 5y ago

   The fundamental idea is "operating in high dimensions", and this does have some solid footing, e.g. see Cover's Theorem ( https://en.wikipedia.org/wiki/Cover's_theorem) . In fact I recently did a pres
3. 243.▲
   by armcat 5y ago

   In the conclusion the authors state: "It would be particularly interesting to see whether such a design works in NLP or other domains." I think it's quite common to use quasi "MLP mixtures" in natural language proce
4. 244.▲
   by armcat 5y ago

   At the start of the pandemic there was little-to-no testing in Sweden, but since autumn 2020 it has expanded and scaled significantly. We have drive-thru tests as well as home-delivered tests, free of charge. I used them a number of times,
5. 245.▲
   by armcat 5y ago

   Faiss looks very nice, thanks for the tip!
6. 246.▲
   by armcat 5y ago

   NN as in "neural network", or NN as in "nearest neighbour" algorithm? No to the former, yes to the latter. The reason for a "no" to neural networks - in my case I've only ever implemented neural networks w
7. 247.▲
   Paper Walkthrough: Exploring Random Encoders for Sentence Classification(github.com)

   2 points by armcat 5y ago|0 comments