NEAL Home Page
Neuromorphic Embodied Agents that Learn
The Neuromorphic Embodied Agents that Learn project (NEAL) is part of the
The Human Brain Project
funded by the EU. This project builds on our earlier
Assembly Robot (CABot)
work and extends it. NEAL will translate the
java based fatiguing leaky integrate and fire neural model and system
to PyNN, which in turn will be translated to the SpiNNaker and HICANN
neuromorphic hardware. The CABot3 agent will run on these platforms
in its virtual environment. It will then be extended to CABot4 with
increased capabilities for learning and new cognitive models. Learning
will include cell assemblies that represent visual object categories
and that persist for psychologically realistic times, and improved
learning of plans.
The project members include:
Talks from the project include:
- A Proto Neural Cognitive
Architecture, at the AI seminar at Middlesex in October 2018
- Big Neural Networks and
Neural Agents, at
the London quarterly meeting in June 2017; notes for a 20 minute talk
(followed by 15 minutes for questions) by Chris.
- NEAL Quarterly Update, at
the London quarterly meeting in June 2017; notes for a 5 minute talk
- NEAL: Milestones and Future
the Dresden quarterly meeting in March 2017; notes for a 6 minute talk
- PlaNeural: Spiking Neural Networks
that Plan, Ian's
BICA 2016 talk.
- FSAs and Agents, at
the Heidelberg quarterly meeting in July 2016; notes for a 10 minute talk
- NEAL and NEAL-2, at
the Paris quarterly meeting in May 2016; notes for a 10 minute talk
- Final NEAL SP9/11 talk, at
the Brighton quarterly meeting in Oct 2015; notes for a 10 minute talk
Natural Language Processing with Neurons that I gave
to the Sheffield NLP group in October 2015.
- The NEAL plan, the kickoff talk
in Sabanaci in April 2014; notes for a 10 minute talk by Chris.
Neuromorphic Embodied Agents that Learn, a talk for the AI reading
group before we submitted the proposal in October 2013.