2016 SR&ED Claim

2016 Scientific Research and Experimental Development tax credit claim

T661 - Part 2 – Project Information

Section B – Project Description

Line 242 -– What scientific or technological uncertainties did you attempt to overcome- uncertainties that could not be removed using standard practice? (Maximum 350 words)

The project is attempting to develop general purpose learning and thinking software for the autonomous control of robots. The approach is based on the development of a new artificial neural network (ANN).

Current robot control software is not general purpose. It is specifically developed to function in a predefined environment with predefined tasks to be performed. An example is Google’s self-driving car control software. Learning capability, if present, is limited to this environment. General purpose artificial intelligence (AGI) architectures exist but are not designed for robot control i.e. they are not grounded on sensory input and device activation. Examples include ACT-R, SOAR, CLARION, LIDA, SIGMA and HTM.  These AGI architectures are also applied only in specifically defined environments. DRAMA, MAXSON and Brook’s COG are experimental architectures designed for robot control but have not been applied with any general purpose success. When such architectures are grounded on real sensory input they also make heavy use of stochastic inference that becomes complex and is a heavy computational overhead. One of the reasons for the complexity of these architectures is that they do not use the same structures for pattern recognition and motor action control.

This research is developing a new architecture to solve these problems. However there are no simple hierarchical (deep learning) ANNs based on binary nodes that are feed-forward and use reinforcement learning. There are also no ANNs that grow nodes in a hierarchy for both recognizing objects and sequences as well as for learning new action habits. Even the deep learning pattern recognition ANNs, such as convolutional neural networks, do not add nodes dynamically as they learn.

More specifically the project is attempting to develop a new hierarchical structure of ANN nodes that:

  • Uses binary neurons (binons) with reinforcement learning.
  • Performs reinforcement learning based on intrinsic motivation.
  • Grows the ANN by adding nodes for parallel (spatial) and sequential (temporal) pattern recognition. Does not separate the training phase from the testing phase – is always learning.
  • Converts graduated sensor readings (sub-symbolic) into symbolic stimuli.
  • Transfers short term memory traces into long term memory.
  • Learns action habits and integrates them into the associated hierarchical behavior network.
  • Executes action habits subconsciously while thinking about other actions (simulation based on experience).

Line 244 – What work did you perform in the tax year to overcome the scientific or technological uncertainties described in Line 242? (Summarize the systematic investigation) (Maximum 700 words)

Approach: Increase the complexity of the ANN structure and algorithms to process the requirements at greater levels of complexity while still handling the lower complexity features. Run regression tests on already working lower complexity features. Determine its success at learning and thinking based on the observation of its actions in artificial test environments simulated in software and by inspection of its internal memory traces and processes. More details about the research are available at www.adaptroninc.com

Two pieces of software are being researched;

A) Morse code is being used for object (spatial) and sequential (temporal) pattern recognition from multi-sensor senses producing graduated (ratio scale) and discrete (symbolic / nominal) stimulus readings using a hierarchy of binons.

B) Hand written digits are being used for object (spatial) pattern recognition using a topological activation structure and binon hierarchy for long term memory of objects and their associated labels.

In area A:

  1. Changed binons to a state and spike-driven model (Idling, Triggering, and Expecting). Binons count the firings of their trigger and goal source binons and fire when they recognize their pattern.
  2. Applied stem binons as precursor nodes before adding learnt binons to long-term memory.
  3. Replaced stem binons with a short-term memory structure that keeps longer term stimuli for higher levels of temporal complexity.

In area B:

  1. Tried using the state and spike driven binon approach from area A to spatial pattern recognition of hand drawn digits.

Line 246 – What scientific or technological advancements did you achieve as a result of the work described in line 244? (Maximum 350 words)

  1. Found out how to propagate spikes / firings of binons up the hierarchy of state- driven binons to recognize repeating temporal patterns at all levels of complexity, not just the sensory level.
  2. Determined that when using the temporal state-driven binon approach to recognize spatial patterns it works but does not perform as well as an approach based on a topographical activation structure. Temporal pattern recognition uses a depth-first process while spatial recognition requires a breadth-first process.