主催: The Japanese Society for Artificial Intelligence
会議名: 2013年度人工知能学会全国大会(第27回)
回次: 27
開催地: 富山県富山市 富山国際会議場
開催日: 2013/06/04 - 2013/06/07
Human beings have been shown to have many kinds of cognitive limits and biases. We generally cannot optimize but just satisfice. Our valuation of an object or option are not absolute but comparative or relative to the others. We have contrasting risk attitude for gain and loss situation. In 2-armed bandit problems, we show that the three limits or biases can, when combined appropriately, work adaptively, actually optimally, and help us to overcome some tradeoffs.