2015/04/14 by Song-Ju Kim, Kim, Song-Ju, Makoto Naruse +3
Computer Science · Decision Sciences · Neuroscience · #Advanced Bandit Algorithms Research #Artificial Intelligence (cs.AI) #FOS: Computer and information sciences #Neural Networks and Reservoir Computing #Neural dynamics and brain function
paper · pdf · doi:10.48550/arxiv.1504.03451
openalex publication_date 2015/04/14 · openalex created_date 2016/06/24 · openalex updated_date 2026/07/28
Each individual handles many tasks of finding the most profitable option from a set of options that stochastically provide rewards. Our society comprises a collection of such individuals, and the society is expected to maximise the total rewards, while the individuals compete for common rewards. Such collective decision making is formulated as the `competitive multi-armed bandit problem (CBP)', requiring a huge computational cost. Herein, we demonstrate a prototype of an analog computer that efficiently solves CBPs by exploiting the physical dynamics of numerous fluids in coupled cylinders. This device enables the maximisation of the total rewards for the society without paying the conventionally required computational cost; this is because the fluids estimate the reward probabilities of the options for the exploitation of past knowledge and generate random fluctuations for the exploration of new knowledge. Our results suggest that to optimise the social rewards, the utilisation of fluid-derived natural fluctuations is more advantageous than applying artificial external fluctuations. Our analog computing scheme is expected to trigger further studies for harnessing the huge computational power of natural phenomena for resolving a wide variety of complex problems in modern information society.