Simulation and implementation of bio-inspired neural networks

Background:

During the last decades, researchers have been trying to study and replicate biological neural systems to understand natural intelligence. Apart from its medical/biological applications, from an information processing viewpoint, it can eventually lead to a new generation of computational devices. Resembling the biological central nervous system, such computational devices open new horizons intelligent, low power, fast, and tolerant to both hardware and data failures.

Brief description:

Building a platform for artificial creatures using a Game Engine (GE). This platform will be used as an environment to develop, simulate, and study the behavior of Spiking Neural Networks (SNN). The platform must be capable of running evolutionary algorithms, i.e. Genetic Algorithm (GA). As a result of simulation, successful creatures (SNN structures) will be selected to be implemented on hardware as a proof of concept.

Goals/Tasks:

• A modular simulation platform based on a game engine.
• SNN-based artificial creatures design for the simulator.
• Basic competition-fitness scenarios for the artificial creatures in the simulator environment.
• Genetic algorithm model and simulation of SNN-based creatures in the simulator.
• Hardware design and HDL simulation of SNN-based creatures.
• Hardware synthesis of the most successful creatures on FPGA.

Required skills:

• Computer programming (C++ or Python depending on the game engine platform).
• Hardware design of SNN and HDL description and simulation.
• Working and design implementation on FPGA.