Forebrain Architectures, Algorithms and Hardware for Cognitive Neuromorphic Agents v2
The goal of FORETHOUGHT is to develop cognitive architectures and algorithms whose form and operation are suitable both for explaining observed biological behavior, as well as for evoking real-time cognitive behaviors in neuromorphic agents.
UPDATE: unfortunately the project proposal was rejected. No comment....
Objective ICT-2011 9.11: FET Proactive: Neuro-Bio-Inspired Systems (NBIS)
Brains are remarkable computing systems which clearly outperform conventional architectures in many real-world tasks. Computational neuroscience has made tremendous progress in uncovering the key principles by which neural systems process information, and ICT has advanced to a point where it is possible to integrate a comparable number of
transistors in a VLSI system as neurons in a mammalian brain. Yet we are still unable to build artificial systems with basic "thinking" abilities comparable with even simple insect brains.
In particular, this objective addresses the need to:
- learn more about the relationship between structure, dynamics and function in neuronal circuits and assemblies, and how information is represented or âcodedâ in a brain.
- develop deeper and more comprehensive theories of neural processing, possibly building on results obtained in the domains of dynamic and complex systems.
- close the gap between neuroscience and engineering by motivating interdisciplinary work that ties data with theories, novel computing paradigms, models and implementations.
a) Developing and applying radically new neural recording, imaging or interfacing concepts and designs for a deeper understanding of neural information processing.
b) New multi-scale dynamical theories of neural representation for the development of neuro-bio-ICT systems that can perform high-level tasks (e.g. robust object recognition, or classification), going beyond purely sensory-driven information processing.
c) Development and prototyping of modular brain-like computing architectures that combine neural processing primitives to give a better understanding of brain function and facilitate the design of more complex processing systems for real-time and optimized performance.
d) World-class global research cooperation and alliances in this area, and links with similar actions outside Europe, in particular with participants from USA and Japan.
IP/STREP proposals should address at least 2 of a), b) or c). CSA proposal should address d).
Target outcome a-c):
â¢ New computing paradigms leading to advanced bio-inspired sensing and processing systems, which are naturally able to learn and adapt
â¢ New concepts leading to new brain-computer interface technologies
â¢ Target outcome d):
â¢ New EU and global collaborations between researchers in multiple disciplines spanning engineering, physical and life science domains.
a-c): IP, STREP
Indicative budget distribution
- IP/STREP: EUR 22 million
- CSA: EUR 1 million