Stochastic rounding and reduced-precision fixed-point arithmetic for solving neural ordinary differential equations

• Published in: Philosophical transactions of the Royal Society of London. Series A: Mathematical, physical, and engineering sciences Vol. 378; no. 2166; p. 20190052
• Main Authors: Hopkins, Michael, Mikaitis, Mantas, Lester, Dave R., Furber, Steve
• Format: Journal Article
• Language: English
• Published: England The Royal Society Publishing 06.03.2020
• Subjects: Action Potentials - physiology; Algorithms; Computer Simulation; Humans; Machine Learning; Models, Neurological; Software; Stochastic Processes; fixed-point arithmetic; stochastic rounding; dither; ordinary differential equation; SpiNNaker; Izhikevich neuron model

Although double-precision floating-point arithmetic currently dominates high-performance computing, there is increasing interest in smaller and simpler arithmetic types. The main reasons are potential improvements in energy efficiency and memory footprint and bandwidth. However, simply switching to...