A One Hot Vector: Difference between revisions
From Algolit
| Line 10: | Line 10: | ||
=one-hot-vectors= | =one-hot-vectors= | ||
| + | <br> | ||
"''Meaning is this illusive thing that were trying to capture''" (Richard Socher in [https://www.youtube.com/watch?v=xhHOL3TNyJs&index=2&list=PLcGUo322oqu9n4i0X3cRJgKyVy7OkDdoi CS224D Lecture 2 - 31st Mar 2016 (Youtube)]) | "''Meaning is this illusive thing that were trying to capture''" (Richard Socher in [https://www.youtube.com/watch?v=xhHOL3TNyJs&index=2&list=PLcGUo322oqu9n4i0X3cRJgKyVy7OkDdoi CS224D Lecture 2 - 31st Mar 2016 (Youtube)]) | ||
| + | <br> | ||
If we work with the example sentence ... | If we work with the example sentence ... | ||
| + | <br> | ||
"The algoliterary explorers discovered a multidimensional landscape made of words disguised as numbers." | "The algoliterary explorers discovered a multidimensional landscape made of words disguised as numbers." | ||
| + | <br> | ||
... these are the 14 words we work with ... | ... these are the 14 words we work with ... | ||
| + | <br> | ||
a | a | ||
algoliterary | algoliterary | ||
| Line 32: | Line 37: | ||
words | words | ||
. | . | ||
| − | + | <br> | |
... a single vector in a one-hot-vector looks like this ... | ... a single vector in a one-hot-vector looks like this ... | ||
| + | <br> | ||
[0 0 0 0 0 0 0 0 0 0 0 0 0 0] | [0 0 0 0 0 0 0 0 0 0 0 0 0 0] | ||
| + | <br> | ||
... and a full fourteen-dimensional matrix like this ... | ... and a full fourteen-dimensional matrix like this ... | ||
| + | <br> | ||
[[0 0 0 0 0 0 0 0 0 0 0 0 0 0] a | [[0 0 0 0 0 0 0 0 0 0 0 0 0 0] a | ||
[0 0 0 0 0 0 0 0 0 0 0 0 0 0] algoliterary | [0 0 0 0 0 0 0 0 0 0 0 0 0 0] algoliterary | ||
| Line 54: | Line 62: | ||
[0 0 0 0 0 0 0 0 0 0 0 0 0 0] words | [0 0 0 0 0 0 0 0 0 0 0 0 0 0] words | ||
[0 0 0 0 0 0 0 0 0 0 0 0 0 0]] . | [0 0 0 0 0 0 0 0 0 0 0 0 0 0]] . | ||
| + | <br> | ||
... with one 0 for each unique word in a vocabulary, and a row for each unique word. | ... with one 0 for each unique word in a vocabulary, and a row for each unique word. | ||
| Line 59: | Line 68: | ||
The following step is to count how often a word appears next to another ... | The following step is to count how often a word appears next to another ... | ||
| + | <br> | ||
"The algoliterary explorers discovered a multidimensional landscape made of words disguised as numbers." | "The algoliterary explorers discovered a multidimensional landscape made of words disguised as numbers." | ||
| + | <br> | ||
| + | <br> | ||
[[0 0 0 1 0 0 0 0 1 0 0 0 0 0] a | [[0 0 0 1 0 0 0 0 1 0 0 0 0 0] a | ||
[0 0 0 0 0 1 0 0 0 0 0 1 0 0] algoliterary | [0 0 0 0 0 1 0 0 0 0 0 1 0 0] algoliterary | ||
| Line 75: | Line 87: | ||
[0 0 0 0 1 0 0 0 0 0 1 0 0 0] words | [0 0 0 0 1 0 0 0 0 0 1 0 0 0] words | ||
[0 0 0 0 0 0 0 0 0 1 0 0 0 0]] . | [0 0 0 0 0 0 0 0 0 1 0 0 0 0]] . | ||
| + | <br> | ||
| + | |||
| + | ==Algolit one-hot-vector scripts== | ||
| + | |||
| + | Two one-hot-vector scripts were created during one of the Algolit sessions, both creating the same matrix but in a different way. To download and run them, use the following links: [https://gitlab.constantvzw.org/algolit/algolit/blob/master/algoliterary_encounter/one-hot-vector/one-hot-vector_gijs.py one-hot-vector_gijs.py] & [https://gitlab.constantvzw.org/algolit/algolit/blob/master/algoliterary_encounter/one-hot-vector/one-hot-vector_hans.py one-hot-vector_hans.py] | ||
=Note that= | =Note that= | ||
Revision as of 15:16, 25 October 2017
| 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))
If we work with the example sentence ...
"The algoliterary explorers discovered a multidimensional landscape made of words disguised as numbers."
... these are the 14 words we work with ...
a algoliterary as discovered disguised explores landscape made multidimensional numbers of the words .
... a single vector in a one-hot-vector looks like this ...
[0 0 0 0 0 0 0 0 0 0 0 0 0 0]
... and a full fourteen-dimensional matrix like this ...
[[0 0 0 0 0 0 0 0 0 0 0 0 0 0] a [0 0 0 0 0 0 0 0 0 0 0 0 0 0] algoliterary [0 0 0 0 0 0 0 0 0 0 0 0 0 0] as [0 0 0 0 0 0 0 0 0 0 0 0 0 0] discovered [0 0 0 0 0 0 0 0 0 0 0 0 0 0] disguised [0 0 0 0 0 0 0 0 0 0 0 0 0 0] explores [0 0 0 0 0 0 0 0 0 0 0 0 0 0] landscape [0 0 0 0 0 0 0 0 0 0 0 0 0 0] made [0 0 0 0 0 0 0 0 0 0 0 0 0 0] multidimensional [0 0 0 0 0 0 0 0 0 0 0 0 0 0] numbers [0 0 0 0 0 0 0 0 0 0 0 0 0 0] of [0 0 0 0 0 0 0 0 0 0 0 0 0 0] the [0 0 0 0 0 0 0 0 0 0 0 0 0 0] words [0 0 0 0 0 0 0 0 0 0 0 0 0 0]] .
... with one 0 for each unique word in a vocabulary, and a row for each unique word.
The following step is to count how often a word appears next to another ...
"The algoliterary explorers discovered a multidimensional landscape made of words disguised as numbers."
[[0 0 0 1 0 0 0 0 1 0 0 0 0 0] a [0 0 0 0 0 1 0 0 0 0 0 1 0 0] algoliterary [0 0 0 0 1 0 0 0 0 1 0 0 0 0] as [1 0 0 0 0 1 0 0 0 0 0 0 0 0] discovered [0 0 1 0 0 0 0 0 0 0 0 0 1 0] disguised [0 1 0 1 0 0 0 0 0 0 0 0 0 0] explores [0 0 0 0 0 0 0 1 1 0 0 0 0 0] landscape [0 0 0 0 0 0 1 0 0 0 1 0 0 0] made [1 0 0 0 0 0 1 0 0 0 0 0 0 0] multidimensional [0 0 1 0 0 0 0 0 0 0 0 0 0 1] numbers [0 0 0 0 0 0 0 1 0 0 0 0 1 0] of [0 1 0 0 0 0 0 0 0 0 0 0 0 0] the [0 0 0 0 1 0 0 0 0 0 1 0 0 0] words [0 0 0 0 0 0 0 0 0 1 0 0 0 0]] .
Algolit one-hot-vector scripts
Two one-hot-vector scripts were created during one of the Algolit sessions, both creating the same matrix but in a different way. To download and run them, use the following links: one-hot-vector_gijs.py & one-hot-vector_hans.py
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, idem.)]
For more notes on this lecture visit http://pad.constantvzw.org/public_pad/neural_networks_3