2019/05/28 by Yassin Khalifa, Khalifa, Yassin, Justin Hawks +3
Computer Science · Materials Science · #FOS: Computer and information sciences #Machine Learning (cs.LG) #Machine Learning (stat.ML) #Machine Learning and ELM #Machine Learning in Materials Science #Neural Networks and Applications
paper · pdf · doi:10.48550/arxiv.1905.12135
openalex publication_date 2019/05/28 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
Recent advances in neuroscience have revealed many principles about neural\nprocessing. In particular, many biological systems were found to\nreconfigure/recruit single neurons to generate multiple kinds of decisions.\nSuch findings have the potential to advance our understanding of the design and\noptimization process of artificial neural networks. Previous work demonstrated\nthat dense neural networks are needed to shape complex decision surfaces\nrequired for AI-level recognition tasks. We investigate the ability to model\nhigh dimensional recognition problems using single or several neurons networks\nthat are relatively easier to train. By employing three datasets, we test the\nuse of a population of single neuron networks in performing multi-class\nrecognition tasks. Surprisingly, we find that sparse networks can be as\nefficient as dense networks in both binary and multi-class tasks. Moreover,\nsingle neuron networks demonstrate superior performance in binary\nclassification scheme and competing results when combined for multi-class\nrecognition.\n