Skip to content

Theano/TensorFlow deep learning programs written while working through various tutorials

Notifications You must be signed in to change notification settings

timestocome/DeepLearning

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Programs written while working through http://deeplearning.net/tutorial/deeplearning.pdf

Original source code for book and data sources are here https://github.com/lisa-lab/DeepLearningTutorials.git

Theano documentation is here: http://deeplearning.net/software/theano/#documentation

Updated list of sources I used to learn machine learning and deep learning: https://github.com/timestocome/Test-stock-prediction-algorithms

This is a place to store the code I write while working through "Deep Learning Tutorial"

My code is in Python 3, book and sample code is Python 2. Not sure which version of Theano he's using, I'm on 0.7/0.8, some minor changes were needed to get his code working running.

I'm also trying to keep each example net self contained instead of broken into separate files like book examples. This is so I can easily drop a network with only the parts I need into a new project. Once that's done it'll be easy to put the various classes back into separate files.

  • see the MachineLearning Resources.md file for a list of tutorials, books etc on Deep Learning

---------- Theano examples ----------

Logistic Regression

Multilayer Perceptron

Simple Multilayer Perceptron ( stripped down version )

Basic Convolutional Network to filter images

LeNet5

Denoising Autoencoders

Stacked Denoising Autoencoders

Restricted Boltzmann Machine

Deep Belief Network

Hybrid Monte-Carlo Sampling

Elman-Recurrent (Python 2.x-3.x and replaced PERL script with slow but working Python) ( clean up code, fix a few bugs )

LSTM for Sentiment Analysis I converted the example code to process a series of ints and predict next int in series instead.

RNN-RBM to model and sequence music ( I changed this to handle chars from reading in plain text )

----- Not in the book: -----

Deep Reinforcement Learning (WIP, the Theano code needs streamlining but it works)

Single layer RNN to read in sentences and generate new sentences

ResNet example of a deep layer residual network, tested on Iris Data set

---------- TensorFlow Examples ----------

Programs written while working through Manning 'Machine Learning with TensorFlow' (https://www.manning.com/books/machine-learning-with-tensorflow )

LogisticRegression_tf ( cluster some generated data with Logistic Regression )

KMeans_tf ( read in dir of .wav files, compute frequency, notes, histograms, cluster with KMeans )

SOM_tf ( Self organizing map that organizes an array of colors )

Reinforcement_tf (Reinforcement ( QLearning) on a single stock )

LSTM_tf (LSTM on airline passenger volume year over year)

Convolutional_tf ( simple convolutional to label images )

Reinforcement Learning ( teach a robot to avoid obstacles https://github.com/timestocome/RaspberryPi/tree/master/ObstacleAvoidance )

----- Not in the book: -----

HelloWorldSeq2Seq.py ( translates Hello to World in TensorFlow )

FullyConnectedForwardFeedNetwork.py ( predict next year's sunspots based on past data in Theano )

NasdaqPrediction.py ( fully connected forward feed w/ regularization for Nasdaq stock index one month out in Theano )

MLP_tf.py ( a very simple MultiLayer Perceptron in TensorFlow )

Residual_1DConvolution.py ( tf 1 layer version of the residual convolutional network described in Ng's paper )

SantaFe_QLearning_tf ( Q learning table on SantaFe Ant Maze )

----- Keras Examples -----

Arrhythmia data Stock GAN, semi-supervised time series GAN

About

Theano/TensorFlow deep learning programs written while working through various tutorials

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published