Skip to content
/ cemer Public

Original C++ version of emergent: originally hosted under svn at https://grey.colorado.edu/svn/emergent

License

GPL-2.0 and 2 other licenses found

Licenses found

GPL-2.0
COPYING
GPL-2.0
COPYING.txt
LGPL-2.1
COPYING.LIB
Notifications You must be signed in to change notification settings

emer/cemer

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

emergent

This is the C++ version of emergent. Main documentation is here: https://grey.colorado.edu/emergent -- basic install and build information is on the Github Wiki

We are no longer developing this software. We are now developing a new framework based on the Go language, with a Python interface: https://github.com/emer/emergent

The current release built from github sources (source code only) is: https://github.com/emer/cemer/releases/tag/v8.6.1

The latest release with built packages is: https://github.com/emer/cemer/releases/tag/v8.5.2, released Feb, 2018.

The main dependency files for current releases are found in: https://github.com/emer/cemer/releases/tag/v8.5.1

This github repository was converted from the svn original, and has captured the full history of git tags (including historical dates!) in: https://github.com/emer/cemer/tags

About

emergent is a comprehensive neural network simulator that enables the creation and analysis of complex, sophisticated models of the brain in the world; features:

  • Full browser and 3D GUI for constructing, visualizing, & interacting.

    • Accessible to non-programmers
    • But also highly productive for experts, used daily in scientific research.
  • Powerful C++ scripting language, css (not ''that'' css), GUI Programming environment (IDE) -- TypeAccess access to C++.

  • Rich, dynamic, embodied environments for training networks:

    • DataTable for network inputs and DataProc, DataAnal, DataGen (filtering, grouping, sorting, dimensionality reduction, graphing, etc).

    • Newtonian physics simulator for Virtual Environment, e.g., a biophysically realistic human arm, and realistic embodied, dynamic vision.

    • Sensory filtering for vision, audition, and vocal-tract speech.

  • Many classic neural network algorithms and variants: Backpropagation (e.g., deep convolutional neural networks), Constraint Satisfaction, Self Organizing, and the Leabra algorithm which incorporates many of the most important features from each of these algorithms, in a biologically consistent manner. Also, symbolic / subsymbolic ACT-R.

    • Highly optimized vector-based back-end code with thread-specific memory allocation, and GPU (CUDA); Convenient compute cluster for GUI-based job control and data management.
  • In use for decades, for hundreds of scientific publications from a variety of different labs. Detailed models of the hippocampus, prefrontal cortex, basal ganglia, visual cortex, cerebellum, etc.

    • Direct descendant of earlier simulators: PDP (1986) and PDP++ (1995).