2019/09/11 by Xueyuan She, She, Xueyuan, Saibal Mukhopadhyay +2
Engineering · Neuroscience · #Advanced Memory and Neural Computing #Emerging Technologies (cs.ET) #FOS: Computer and information sciences #Ferroelectric and Negative Capacitance Devices #Neural and Evolutionary Computing (cs.NE) #Neural dynamics and brain function
paper · pdf · doi:10.48550/arxiv.1909.05401
openalex publication_date 2019/09/11 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
Spike-timing-dependent-plasticity (STDP) is an unsupervised learning\nalgorithm for spiking neural network (SNN), which promises to achieve deeper\nunderstanding of human brain and more powerful artificial intelligence. While\nconventional computing system fails to simulate SNN efficiently,\nprocess-in-memory (PIM) based on devices such as ReRAM can be used in designing\nfast and efficient STDP based SNN accelerators, as it operates in high\nresemblance with biological neural network. However, the real-life\nimplementation of such design still suffers from impact of input noise and\ndevice variation. In this work, we present a novel stochastic STDP algorithm\nthat uses spiking frequency information to dynamically adjust synaptic\nbehavior. The algorithm is tested in pattern recognition task with noisy input\nand shows accuracy improvement over deterministic STDP. In addition, we show\nthat the new algorithm can be used for designing a robust ReRAM based SNN\naccelerator that has strong resilience to device variation.\n