2017/11/09 by Shruti Kulkarni, Kulkarni, Shruti R., John M. Alexiades +3
Engineering · Neuroscience · #Advanced Memory and Neural Computing #Artificial Intelligence (cs.AI) #FOS: Computer and information sciences #Ferroelectric and Negative Capacitance Devices #Machine Learning (cs.LG) #Machine Learning (stat.ML) #Neural and Evolutionary Computing (cs.NE) #Neural dynamics and brain function
paper · pdf · doi:10.48550/arxiv.1711.03637
openalex publication_date 2017/11/09 · openalex created_date 2022/10/03 · openalex updated_date 2026/07/28
We describe a novel spiking neural network (SNN) for automated, real-time\nhandwritten digit classification and its implementation on a GP-GPU platform.\nInformation processing within the network, from feature extraction to\nclassification is implemented by mimicking the basic aspects of neuronal spike\ninitiation and propagation in the brain. The feature extraction layer of the\nSNN uses fixed synaptic weight maps to extract the key features of the image\nand the classifier layer uses the recently developed NormAD approximate\ngradient descent based supervised learning algorithm for spiking neural\nnetworks to adjust the synaptic weights. On the standard MNIST database images\nof handwritten digits, our network achieves an accuracy of 99.80% on the\ntraining set and 98.06% on the test set, with nearly 7x fewer parameters\ncompared to the state-of-the-art spiking networks. We further use this network\nin a GPU based user-interface system demonstrating real-time SNN simulation to\ninfer digits written by different users. On a test set of 500 such images, this\nreal-time platform achieves an accuracy exceeding 97% while making a prediction\nwithin an SNN emulation time of less than 100ms.\n