https://github.com/ashenoy95/ml

Machine Learning, Neural Nets, Deep Learning basics
https://github.com/ashenoy95/ml

backpropagation expectation-maximization-algorithm gmm lstm ml mlp mnist-classification neural-network python3 rnn-tensorflow speech-denoising

Last synced: about 1 month ago
JSON representation

Machine Learning, Neural Nets, Deep Learning basics

• Host: GitHub
• URL: https://github.com/ashenoy95/ml
• Owner: ashenoy95
• Created: 2017-04-23T08:41:28.000Z (over 7 years ago)
• Default Branch: master
• Last Pushed: 2018-04-28T07:56:15.000Z (over 6 years ago)
• Last Synced: 2024-10-27T14:44:38.012Z (about 2 months ago)
• Topics: backpropagation, expectation-maximization-algorithm, gmm, lstm, ml, mlp, mnist-classification, neural-network, python3, rnn-tensorflow, speech-denoising
• Language: Jupyter Notebook
• Homepage:
• Size: 4.36 MB
• Stars: 1
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# Machine Learning

* __speech-denoising__: MLP and RNN based speech denoiser implementations using TensorFlow.

* __em.py__: EM algorithm for a Mixture of Gaussians on one-dimensional data.

* __mnist-shallow.py__: TensorFlow implementation of MLP (1024x5) using ReLU activation, He initialization and Adam optimization giving over 98% accuracy. Also analysing TSNE & PCA. 

* __nw_compression.ipynb__: Network compression of 'mnist-shallow' using low rank approximation of trained weights (using the top 20 singular values after SVD), thereby using only about 4% of the memory of the original network.

* __parity.py__: MLP (4-4-1) with backprop from scratch to solve the parity problem for a 4-bit input.