Yoshiyuki Kotani

Learn More
This paper presents a new reinforcement learning method, called temporal difference learning with Monte Carlo simulation (TDMC), which uses a combination of Temporal Difference Learning (TD) and winning probability in each non-terminal position. Studies on self-teaching evaluation functions as applied to logic games have been conducted for many years,(More)
We describe a corpus-based approach of natural language dialogue system. The characteristic is that all the system's behaviors, like processing and understanding dialogues and generating responses, depend on corpora. As a result, the system can handle any language and any topic. This paper aims to explain the whole architecture and individual technology(More)
Introduction. According to the analysis of grandmaster-like strategies in Shogi [Iida and Uiterwijk 1993], it is important for a teacher, at the beginning stages of teaching, to intentionally lose an occasional game against a novice opponent, or to play less than optimally in order to give the novice some prospects of winning, without this being noticed by(More)
— Nested Monte-Carlo Search, which calls Monte-Carlo search in the nested call, has succeeded in the one-person game named Morpion Solitaire. The depth for the nest is called a level, and the runtime increases exponentially in the search for higher level. In the present study, All-Move-As-First heuristic is incorporated in Nested Monte-Carlo Search and the(More)
Japanese has thousands of onomatopoeias and they have recently started to attract a lot of attention of researchers on natural language processing. Some onomatopoeias are semantically or phonologically similar each other and the choice of these onomatopoeias sometimes give a big difference among Japanese sentences. In this paper, the authors classify(More)
Monte-Carlo method recently has produced good results in Go. Monte-Carlo Go uses a move which has the highest mean value of either winning percentage or final score. In a past research, winning percentage is superior to final score in Monte-Carlo Go. We investigated them in BlokusDuo, which is a relatively new game, and showed that Monte-Carlo using final(More)
Typically, studies of ubiquitous computing environments focus on context-aware assistance for activities in a closed domain, such as guide systems in museums, and ignore ways of coordinating services between two different service environments. However, it is important to provide them with personalized assistance services that seamlessly coordinate two(More)
MTD algorithm developed by Plaat is a variation of SSS* which uses the depth-first strategy to resolve the storage problem coming from the best-first strategy. Since MTD algorithm is based on the zero window search algorithm, the initial range of the searching windows plays an important role in the performance. In this paper, we show some basic experimental(More)