Neural-Mass model acceleration on the ACAP heterogeneous computing platform [Theme: BrainFrame]

Mean-field approximation of network of biophysical neurons driven by  conductance-based ion exchange | bioRxiv

The neural-mass modeling problem can be stated in generalized form using ANNs (in the form of neural ODEs) on the nodes with sparse (delayed) coupling in between. In this form, numerical accuracy matters less and the problem is amenable for fixed-point-based simulation. The ACAP platform contains an FPGA fabric, powerful AI engines (i.e., vector processors) and a high-bandwidth and low-latency on-chip interconnect between them. With node dynamics running on the AI cores, edge dynamics on the FPGA, plenty of on-chip memory and memory-intensive delay coupling scheduled via DRAM, this could possibly be the perfect platform to run neural-mass models.

Job Category: MSc Thesis topic
Job Type: Student Thesis

Apply for this position

Allowed Type(s): .pdf, .doc, .docx