November 2016 – Ogma is pleased to announce the release of its OGMANEO software library, alongside usage examples and demonstrations, and language bindings for Java™ and Python™. The OgmaNeo library contains implementations of Online Predictive Hierarchies, as described in the paper: “Feynman Machine: The Universal Dynamical Systems Computer” (http://arxiv.org/abs/1609.03971). In September, our paper was listed as among the Best of the Physics arXiv by MIT Technology Review.
Feynman Machine Abstract
Efforts at understanding the computational processes in the brain have met with limited success, despite their importance and potential uses in building intelligent machines. We propose a simple new model which draws on recent findings in Neuroscience and the Applied Mathematics of interacting Dynamical Systems. The Feynman Machine is a Universal Computer for Dynamical Systems, analogous to the Turing Machine for symbolic computing, but with several important differences. We demonstrate that networks and hierarchies of simple interacting Dynamical Systems, each adaptively learning to forecast its evolution, are capable of automatically building sensorimotor models of the external and internal world. We identify such networks in mammalian neocortex, and show how existing theories of cortical computation combine with our model to explain the power and flexibility of mammalian intelligence. These findings lead directly to new architectures for machine intelligence. A suite of software implementations has been built based on these principles, and applied to a number of spatiotemporal learning tasks.
The OgmaNeo software library
The C++ library is available for use on Microsoft® Windows®, Apple® OS X®, Raspberry Pi® Jessie, and Linux® operating systems. It uses the OpenCL™ 1.2 (Open Computing Language) framework to enable execution across heterogeneous platforms.
The library is now available on the Ogma GITHUB® website – https://github.com/ogmacorp It is published under the Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International (CC BY-NC-SA 4.0) License.
Building the library requires: a C++ 11 compiler, an OpenCL Software Development Kit, the CMake™ build framework, and uses the Google™ FlatBuffers package for fast and memory-efficient data serialization.
OgmaNeo demonstrations perform state of the art online learning for:
- Video Prediction,
- Anomaly Detection,
- Sequence and Grammar Prediction.
The Ogma YouTube™ channel contains videos that show video prediction and anomaly detection – https://www.youtube.com/ogmaai
Ogma is building new AI technology based on a multidisciplinary approach, combining the latest developments in machine learning, the applied mathematics of dynamical systems, and computational neuroscience. Systems matching the power of state-of-the-art Deep Learning algorithms can be built with the self-organising structure, flexibility and efficiency found in the human brain.
For further information visit https://ogma.ai.
Copyright © 2016 Ogma Intelligent Systems Corp.