https://github.com/azimjohn/language-classifier
Language Classifier using single layer network of perceptrons
https://github.com/azimjohn/language-classifier
Last synced: over 1 year ago
JSON representation
Language Classifier using single layer network of perceptrons
- Host: GitHub
- URL: https://github.com/azimjohn/language-classifier
- Owner: azimjohn
- Created: 2022-04-07T11:18:53.000Z (over 4 years ago)
- Default Branch: master
- Last Pushed: 2022-11-03T20:40:32.000Z (over 3 years ago)
- Last Synced: 2023-08-29T16:41:45.604Z (almost 3 years ago)
- Language: Python
- Size: 1.03 MB
- Stars: 9
- Watchers: 1
- Forks: 2
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# Language Classifier
## Using Single Layer Network of Perceptrons
The aim of the project is to create a single-layer network to identify the language an input text is written in.
Files contains a set of texts written in four languages ā English, Polish, German and Uzbek. To classify a given text, it counts the number of occurences of each letter of the latin alphabet. Ignores all other characters ā only counts the frequencies of the 26 letters of the latin alphabet.
For each text, it generates a 26-element input vector containing the number of occurences of each letter. Then normalize it:
```
vĖ = v / |v|
```
The output of the network has local representation: one neuron is assigned to each language. For a given text only the appropriate neuron should have value 1 and all others 0. To classify an input text, it selects the neuron with the maximum activation.