https://github.com/pointlander/entity

Multivariate gaussian neural network
https://github.com/pointlander/entity

multivariate-gaussian neural-network

Last synced: 5 months ago
JSON representation

Multivariate gaussian neural network

• Host: GitHub
• URL: https://github.com/pointlander/entity
• Owner: pointlander
• License: bsd-3-clause
• Created: 2025-05-28T01:03:08.000Z (about 1 year ago)
• Default Branch: main
• Last Pushed: 2025-07-25T18:26:17.000Z (11 months ago)
• Last Synced: 2025-07-25T23:11:48.746Z (11 months ago)
• Topics: multivariate-gaussian, neural-network
• Language: Go
• Homepage:
• Size: 1.55 MB
• Stars: 0
• Watchers: 0
• Forks: 0
• Open Issues: 1
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

          
# A multivariate gaussian neural network

This repo implements a [multivariate gaussian](https://en.wikipedia.org/wiki/Multivariate_normal_distribution) 

neural network and works by seeing how close in terms of the L2 norm the samples are to a query vector. 

To generate the distributions the average vector is first calculated, and then the covariance matrix is produced. 

A standard deviation matrix is produced by taking the sqaure root of the covariance matrix. The average 

vector and standard deviation matrix form the model.

## Implementation of the iris data set

The iris data set has 4 feature per flower with three types of flowers. The data set is divided up in terms of 

flower type creating 3 sets with 50 samples each. The 3 sets of samples are used to create 3 multivariate 

gaussians. To classify a query vector of 4 features the multivariate gaussian is sampled from. The L2 norm is 

used to calculate which multivariate guassian produces samples closest to the query vector.