著者
松田 潤
出版者
成蹊大学アジア太平洋研究センター
雑誌
アジア太平洋研究 = Review of Asian and Pacific studies (ISSN:09138439)
巻号頁・発行日
no.41, pp.45-61, 2016

In this paper, the topics concerning of mourning and melancholy are discussed in reference to "Namaeyotattearuke Nakaya Kokichi Ikousyu"(1972). First, the potential and limit of Nakaya's participation in Movement at Ryukyu University though 1960s which was triggered by the death of his niece in the incident of 1959 Okinawa F-100 Crash will be examined in a historical context. I will further examine that the movement was aiming to overcome an aspect of its sectionality and inflexibility and to criticize the significance of Reversion Movement in Japan in the context of historical trend in Neo-leftism. Secondly, Nakaya's journey for 40 days in his "motherland Japan" was the second trigger after the death of his niece to change his perspectives. Nakaya's arrest for occupation in Legislature of the Government of the Ryukyu Islands was evaluated to be separation from representative politics and to be in the process of population becoming "people". Lastly, the poem of Nakaya's which was believed to be Nakaya's fail in political conflicts should be dealt. Nakaya endeavored to perpetuate his poem and death by the means of opposition to the contemporary political system. In addition, whether the way of incorporation of the trace of alterity and endless mourning work or not should be.
著者
松野 陽一郎 山崎 達也 松田 潤 石井 信
出版者
一般社団法人電子情報通信学会
雑誌
電子情報通信学会技術研究報告. NC, ニューロコンピューティング (ISSN:09135685)
巻号頁・発行日
vol.100, no.688, pp.91-98, 2001-03-16

本研究では, マルチエージェント競合系の例としてカードゲームであるハーツを取り上げ, Actor-Cmicアルゴリズムと先読みを用いた強化学習法を提案する. この系では, 部分観測マルコフ決定過程における強化学習を取り扱わねばならない. 提案手法では, ゲームを序盤, 中盤, 終盤の3つの局面に分割し, それぞれの局面でActorを切り換えて行動制御と学習を行う. 中盤においては, 期待TD誤差に基づく先読みを用いた行動選択を行う. この際, Criticによって近似された状態評価関数と相手戦略から推定した状態遷移確率を用いて期待TD誤差を計算することで部分観測性に対処する. 本手法の有効性を計算機シミュレーション実験により確認した.