kaikun213 / uni-ht17-comp3211-htm-mnist

COMP3211 course project. Very simple handwriting recognition algorithm.

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

MNIST Recognition

HTM vision algorithm used for basic digit recognition of the MNIST Dataset.

  • Only small course project. Report will be added later.
  • Small changes to the code to work with current nupic version.
  • Proven Network parameters adopted, no extra swarming done.
  • Further work should integrate features for better results. (Currently ~95% accuracy as in original vision repo):
    • temporal pooling,
    • sequence memory,
    • location-paired-information
    • a local receptive field

Dependencies

Uses NuPIC Vision Toolkit

"This tool kit is intended to help people who are interested in using Numenta's NuPIC on vision problems."

Follow the instructions for installing htmresearch, which requires nupic and nupic.bindings (from nupic.research.core), and then use pip to install nupic.vision.

Install

You need to be connected to the local mysql DB (default: root and no password, config in nupic/src/support/nupic-default.xml) and simply run setup.py

Rest are mentioned in requirements.txt/Dependencies.md

Then follow instructions in src/nupic/vision/mnist to execute the experiments.

About

COMP3211 course project. Very simple handwriting recognition algorithm.

License:GNU Affero General Public License v3.0


Languages

Language:Python 98.8%Language:TeX 0.9%Language:C++ 0.3%Language:Makefile 0.0%Language:Shell 0.0%