https://github.com/calclavia/htm-tensorflow

An implementation of Numenta's HTM algorithm in Tensorflow.
https://github.com/calclavia/htm-tensorflow

Last synced: 3 months ago
JSON representation

An implementation of Numenta's HTM algorithm in Tensorflow.

• Host: GitHub
• URL: https://github.com/calclavia/htm-tensorflow
• Owner: calclavia
• Created: 2016-11-04T19:03:40.000Z (about 8 years ago)
• Default Branch: master
• Last Pushed: 2019-02-13T06:57:11.000Z (over 5 years ago)
• Last Synced: 2024-07-19T03:04:58.097Z (4 months ago)
• Language: Python
• Homepage:
• Size: 43 KB
• Stars: 67
• Watchers: 13
• Forks: 20
• Open Issues: 4
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# Hierarchical Temporal Memory in Tensorflow

An implementation of Numenta's HTM algorithm in Tensorflow with GPU support.

API design based on Keras API.

## Setup

Install Python 3.5 and PIP. Then run the following command to install all project

dependencies.

```

pip install -r requirements.txt

```

See Tensorflow's documentation on GPU setup.

## Experiments

### MNIST

Experiment with MNIST dataset using an HTML spatial pooler and 1 layer neural

network softmax classifier.

Ensure that the MNIST dataset is placed into the data folder in its zipped format.

http://yann.lecun.com/exdb/mnist/

```

python mnist.py

```

Results using the provided hyperparameters achieve ~95% validation accuracy.