Neuromorphic Architecture

Neuromorphic architectures are hardware architectures that use the biologically inspired neural functions as the basis of operation. Information processing based on spiking neuron architectures have caught considerable attention in recent years due to its low power consumption compared to traditional artificial neural networks.

In this project, as the first stage, we are implementing parallel multiple processing elements based on RISC-V architecture to represent biological neurons. Single neurons can be implemented as a single processor with local memory access or since the spike time of biological neurons is in the millisecond order multiple neurons can be virtualized to a single processor. At the second stage of the process, we are expecting to design encoders and decoders to benchmark the architecture by solving classical machine learning problems.

Team

• E/16/088, Dissanayake D M T H, email, GitHub
• E/16/276, Perera G K B H, email, GitHub
• E/16/366, Thilarakathna W M D U, email, GitHub

Supervisors

• Dr. Isuru Navinna, email
• Dr. Mahanama wickramasinghe, email
• Prof. Roshan G. Ragel, email
• Dr. D.M.I.S. Dasanayake, email

External Links

• Department of Computer Engineering
• Faculty of Engineering
• University of Peradeniya