A One Hot Vector

Type:	Algoliterary exploration
Technique:	word-embeddings
Developed by:	Algolit

one-hot-vectors

"Meaning is this illusive thing that were trying to capture" (Richard Socher in CS224D Lecture 2 - 31st Mar 2016 (Youtube))

[0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0]

with one 0 for each word in a vocabulary where the 1 is representing the place of a word in the vector > In this kind of vector representation: none of the words are similar, they are all a 1.

Note that

Words are represented once in a vector. So words with multiple meanings, like "bank", are more difficult to represent. There is research to multivectors for one word, so that it does not end up in the middle. - Richard Socher, for more notes on this lecture visit http://pad.constantvzw.org/public_pad/neural_networks_3