2017/06/30 by Ahmadreza Ahmadi, Ahmadi, Ahmadreza, Jun Tani +1
Biochemistry, Genetics and Molecular Biology · Computer Science · Neuroscience · #Artificial Intelligence (cs.AI) #Cell Image Analysis Techniques #FOS: Computer and information sciences #Machine Learning (cs.LG) #Neural Networks and Applications #Neural dynamics and brain function
paper · pdf · doi:10.48550/arxiv.1706.10240
openalex publication_date 2017/06/30 · openalex created_date 2022/10/04 · openalex updated_date 2026/07/28
The current paper proposes a novel variational Bayes predictive coding RNN\nmodel, which can learn to generate fluctuated temporal patterns from exemplars.\nThe model learns to maximize the lower bound of the weighted sum of the\nregularization and reconstruction error terms. We examined how this weighting\ncan affect development of different types of information processing while\nlearning fluctuated temporal patterns. Simulation results show that strong\nweighting of the reconstruction term causes the development of deterministic\nchaos for imitating the randomness observed in target sequences, while strong\nweighting of the regularization term causes the development of stochastic\ndynamics imitating probabilistic processes observed in targets. Moreover,\nresults indicate that the most generalized learning emerges between these two\nextremes. The paper concludes with implications in terms of the underlying\nneuronal mechanisms for autism spectrum disorder and for free action.\n