Kim, Guhyun Kornijcuk, Vladimir Kim, Dohun Kim, Inho Kim, Jaewook Woo, Hyo Cheon Kim, Jihun Hwang, Cheol Seong Jeong, Doo Seok 2024-01-19T21:03:00Z 2024-01-19T21:03:00Z 2021-09-05 2019-01 2169-3536 https://pubs.kist.re.kr/handle/201004/120512 In spite of remarkable progress in machine learning techniques, the state-of-the-art machine learning algorithms often keep machines from real-time learning (online learning) due, in part, to computational complexity in parameter optimization. As an alternative, a learning algorithm to train a memory in real time is proposed, named the Markov chain Hebbian learning algorithm. The algorithm pursues efficient use in memory during training in that: 1) the weight matrix has ternary elements (-1, 0, 1) and 2) each update follows a Markov chain-the upcoming update does not need past weight values. The algorithm was verified by two proof-of-concept tasks: image (MNIST and CIFAR-10 datasets) recognition and multiplication table memorization. Particularly, the latter bases multiplication arithmetic on memory, which may be analogous to humans' mental arithmetic. The memory-based multiplication arithmetic feasibly offers the basis of factorization, supporting novel insight into memory-based arithmetic. English IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC SINGLE NEURONS DEEP MEMORY NETWORKS Markov Chain Hebbian Learning Algorithm With Ternary Synaptic Units Article 10.1109/ACCESS.2018.2890543 1 IEEE ACCESS, v.7, pp.10208 - 10223 IEEE ACCESS 7 10208 10223 scie scopus 000457963100001 2-s2.0-85061094324 Computer Science, Information Systems Engineering, Electrical & Electronic Telecommunications Computer Science Engineering Telecommunications Article SINGLE NEURONS DEEP MEMORY NETWORKS Greedy edge-wise training Hebbian learning Markov chain mental arithmetic prime factorization supervised learning ternary unit