Skip to content

Using a fully connected bipartite graph, with no adjacent connections, to create a Restricted Boltzmann Machine we can achieve a 90.0% accuracy on MNIST and 74.5% on Fashion_MNIST

Notifications You must be signed in to change notification settings

siddhant1999/Restricted-Boltzmann-Machine

Repository files navigation

Restricted-Boltzmann-Machine

Using a fully connected bipartite graph, with no adjacent connections, to create a Restricted Boltzmann Machine we can achieve a 90.0% accuracy on MNIST and 74.5% on Fashion_MNIST We also used this to perform Lung Cancer classification

alt text

neuralnet.py is a neural network from scrath (unrelated to RBMs, but can be used as a classifier)

run with nvblas python for faster runtimes (with numpy matrix multiplication)

About

Using a fully connected bipartite graph, with no adjacent connections, to create a Restricted Boltzmann Machine we can achieve a 90.0% accuracy on MNIST and 74.5% on Fashion_MNIST

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages