Getting Started

For a high-level introduction to what NuPIC is and why it exists, see this YouTube video of our presentation at OSCON from July, 2013.

Getting the code (read-only)

• Follow the Github instructions to set up git.
• Clone the repository: git clone https://github.com/numenta/nupic.git

Getting the code (for developers)

• Create a Github account if you don't have one already.
• Install git.
• Generate a key pair for Github
• Follow the Github instructions to fork and clone the repository and set up an upstream remote.

NuPIC Installation

See the installation instructions.

Available howtos for Running NuPIC on:

• Linux
• Mac OSX
• Raspberry Pi
• a virtual machine
• Amazon EC2

Installation FAQ

Build failure - cannot find glibtoolize

Make sure libtoolize is installed. On OS X, simply run brew install libtoolize.

Next Steps

• Read about the Online Prediction Framework
• Look through the simple model code in $NUPIC/examples/opf/clients/hotgym/hotgym.py
• Check out other uses:
  • Linguist - Learns text and predicts upcoming letters.
  • Spatial classification is a classification/clustering task where we classify (=assign labels) to data (one or more input fields)
  • Temporal classification here time plays a role. We try to say "the sequence A-B-C-D looks like LabelABCDSequence"
  • Prediction to predict next (nth step) value in future.
  • Anomaly detection to say how weird/unexpected your input sample is.
  • just create a SDR of the data.