Skip to content

weka511/learn

Repository files navigation

learn

Experiments in machine learning

Projects

File Remarks
Autoencoder Use of autoencoder for training

General

File Remarks
ae.py Simple autoencoder tutorial
gibbs.py A generic Gibbs sampler, based on gibbs.py
hopfield.py Hopfield network
learn.bib Bibliography
learn.wpr Project file for Wing IDE

Imported from elsewhere

File Remarks
bayes1.py Simple demo for pymc3-Code from Estimating Probabilities with Bayesian Modeling in Python
gibbs.py Bayesian Inference: Gibbs Sampling
mhast.py Bayesian Inference: Metropolis-Hastings
naive.py Code from How to Develop a Naive Bayes Classifier from Scratch in Python
UGMM.py CAVI code snarfed from Zhiya Zuo

Keras and Tensorflow explorations

File Remarks
LeNet5.py LeNet-5 CNN in keras
losses.R Plot loss and accuracy for Training and Validation data from logfiles
tf1.py Tensorflow Tutorial
tf2.py Modification of tf1 to use Convolutional layer

Pytorch Learnings

File Remarks
torch-nn.py train

Variational Inference

Programs written to understand Variational Inference, based on the following references:

File Remarks
CAVI.tex Doco for VI programs
cavi1.py CAVI for Univariate Gaussian from Univariate Gaussian Example
cavi3.py The Coordinate Ascent Mean-Field Variational Inference (CAVI) example from Section 3 of Blei et al
cavi.py The Coordinate Ascent Mean-Field Variational Inference (CAVI) example from Section 3 of Blei et al
em.py Expectation Maximization
gmm.py Generate data in accordance with Gaussian Mixture Model
motifs.py Gibbs sampler for finding motifs--Implement GibbsSampler

Free Energy

Programs based on A tutorial on the free-energy framework for modelling perception and learning, by Rafal Bogacz

File Remarks
feex1.py Exercise 1--posterior probabilities
feex2.py Exercise 2--most likely size
feex3.py Exercise 3--neural implementation
feex5.py Exercise 5--learn variance

About

Experiments in machine learning

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published