博士（理学）, 東北大学

Game Informatics

ゲーム情報学

Informatics, Entertainment and game informatics

Informatics, Intelligent informatics

01 Apr. 2015
Graduate School of Informatics and Engineering, Associate Professor

01 Aug. 2010 - 31 Mar. 2015
Center for Frontier Science and Engineering, Assistant Professor

01 Apr. 2010 - 31 Jul. 2010
Center for the Advancement of Higher Education, Tohoku University, Assistant Professor

01 Apr. 2007 - 31 Mar. 2010
Graduate School of Science and Faculty of Science, Tohoku University, Assistant Professor

01 Jun. 2006 - 31 Mar. 2007
Graduate School of Science and Faculty of Science, Tohoku University, Research assistant

01 May 2003 - 31 May 2006
Chemistry Department at University of Toronto, Postdoctoral Research Fellow

01 Apr. 2000 - 31 Mar. 2003
Graduate School of Science and Faculty of Science, Tohoku University, Research Fellow of the Japan Society for the Promotion (DC1)

Apr. 2015
情報処理学会ゲーム情報学運営委員, Society

Apr. 2015
情報処理学会論文誌編集委員, Society

Nov. 2022
情報処理学会ゲーム情報学研究会
NOCCA×NOCCAの強解決
GPW優秀論文賞, 山本敦也;保木邦仁

Nov. 2020
第1回世界将棋AI電竜戦（2020年11月）
第1回世界将棋AI電竜戦独創賞
Others

Jul. 2018
Computer Entertainment Developers Conference (CEDEC)
著述賞
CEDEC AWARDS 2018 特別賞, 伊藤毅志;保木邦仁;三宅陽一郎
Japan society, Japan

Mar. 2015
Information Processing Society of Japan (IPSJ)
長尾真記念特別賞
Japan society

Apr. 2014
International Computer Games Association (ICGA)
ICGA Journal Award
International society

May 2013
Computer Shogi Association (CSA)
第23回世界コンピュータ将棋選手権優勝

Nov. 2012
Taiwanese Association for Artificial Intelligence
Excellent Paper Award
International society

Nov. 2012
Taiwanese Association for Artificial Intelligence
Excellent Paper Award (TAAI 2012)
International society

Jan. 2008
情報処理学会
ソフトウエアジャパン2008アワード

May 2007
Computer Shogi Association
CSA貢献賞

May 2006
Computer Shogi Association (CSA)
第16回世界コンピュータ将棋選手権優勝

Mar. 2003
青葉理学振興会
平成14年度青葉理学振興会賞

NOCCA × NOCCAの強解決
山本敦也; 保木邦仁
Last, 情報処理学会論文誌, 64, 12, 1678-1688, Dec. 2023, Peer-reviwed
Scientific journal, English

Shift Quality Classifier Using Deep Neural Networks on Small Data with Dropout and Semi-Supervised Learning
Takefumi KAWAKAMI; Takanori IDE; Kunihito HOKI; Masakazu MURAMATSU
IEICE Transactions on Information and Systems, Institute of Electronics, Information and Communications Engineers (IEICE), E106.D, 12, 2078-2084, 01 Dec. 2023, Peer-reviwed
Scientific journal, English

Method for constructing artificial intelligence player with abstractions to Markov Decision processes in multiplayer game of Mahjong
Moyuru Kurita; Kunihito Hoki
IEEE Transactions on Games, 13, 1, 99-110, Mar. 2021, Peer-reviwed
Scientific journal, English

Development of Artificial Intelligence to Classify Quality of Transmission Shift Control Using Deep Convolutional Neural Networks
Takefumi Kawakami; Takanori Ide; Eiji Moriyama; Kunihiro Hoki; Masakazu Muramatsu
IEEE Transactions on Vehicular Technology, 69, 12, 16168-16172, Dec. 2020, Peer-reviwed
Scientific journal, English

Recognizing Similarities in Automatic Transmissions of Vehicles by Using Time Series Data and Autoencorders
Takefumi Kawakami; Takanori Ide; Kiyohisa Tomita; Eiji Moriyama; Kunihito Hoki; Masakazu Muramatsu
WCX SAE World Congress Experience, 2019-01-0343, Apr. 2019, Peer-reviwed
International conference proceedings, English

Reinforcement Learning of Elementary Action Knowledge of Digital Curling
Ryohei Matsui; Kunihito Hoki
IPSJ Journal, Information Processing Society of Japan, 59, 11, 2063-2073, 15 Nov. 2018, Peer-reviwed
Scientific journal, Japanese

Classification of Time Series Measurement Data for Lock-Up Clutch of Automatic Transmission of Vehicles Using Deep Convolutional Neural Networks
Takefumi Kawakami; Takanori Ide; Kiyohisa Tomita; Eiji Moriyama; Hiroshi Tsutsui; Kunihito Hoki; Masakazu Muramatsu
SAE Technical Papers, SAE International, 2018-, 2018, Peer-reviwed, A newly developed classifying method for time series measurement data of automatic transmission of vehicles is presented. The proposed method uses deep convolutional neural networks to learn what is comfortable acceleration. In addition, our proposal method employs supervised learning based on experienced engineer's criterion. As a demonstrative problem, we consider the classification of time series measurement data for lock-up clutch control of an 8 speed automatic transmission.
International conference proceedings, English

Improved simulation adjusting
Nobuo Araki; Kunihito Hoki; Masakazu Muramatsu
ICGA Journal, International Computer Games Association, 39, 3-4, 195-204, 2017, Peer-reviwed
Scientific journal, English

Pure Nash equilibria of competitive diffusion process on toroidal grid graphs
Yuki Sukenari; Kunihito Hoki; Satoshi Takahashi; Masakazu Muramatsu
DISCRETE APPLIED MATHEMATICS, ELSEVIER SCIENCE BV, 215, C, 31-40, Dec. 2016, Peer-reviwed, We consider a competitive diffusion process for two players on two-dimensional toroidal grid graphs. Pure Nash equilibria for the game are completely characterized; i.e., for arbitrary-sized toroidal grid graphs, we list all the initial positions that are pure Nash equilibria. (C) 2016 Elsevier B.V. All rights reserved.
Scientific journal, English

畳み込みニューラルネットワークを用いた囲碁における 畳み込みニューラルネットワークを用いた囲碁における一局の棋譜からの棋力推定
荒木伸夫; 保木邦仁; 村松正和
情報処理学会論文誌, 57, 11, 2365-2373, 2016, Peer-reviwed, 囲碁のプロ棋士は，1局の棋譜を見ればプレイヤの棋力が分かるといわれている．本稿では，畳み込みニューラルネットワーク（Convolutional Neural Network; CNN）を使用し，1局の囲碁の棋譜より，プレイヤの棋力を推定する手法を提案する．プレイヤのレート値を推定する実験と，プレイヤを上級/中級/初級にクラス分けする実験を行った．提案手法を実装して囲碁クエストの13路盤棋譜データを用いて学習させて実験したところ，レート値を推定する手法としては従来手法より平均自乗誤差が小さくなった．また，クラス分類する実験においては，1度CNNを用いてレート値を推定してからその値に応じてクラス分けを行う手法と，最初からクラス分類をCNNに学習させる手法の2種類を提案し，それぞれ長所と短所があることを確かめた．It is said that any professional player can estimate a player's strength accurately by looking at just one game record. We propose to use Convolutional Neural Network (CNN) to estimate a Go player's strength from only one game record. We perform two experiments: (i) to estimate a player's rating, and (ii) to classify a player into three classes in strength. We use game records provided by GoQuest to train CNN. For estimating ratings, we compare our method with an existing method to find that our method gives a smaller average mean squared error than that of the existing method. For the classification, we compare two methods: (i) the method that classify a player according to the rating predicted by the CNN, and (ii) the method that trains CNN directly to classify a player based on just one game record. We observed that the two methods have different strong points and weak points.
Scientific journal, Japanese

Human-Like Build-Order Management in StarCraft to Win against Specific Opponent's Strategies
Hiroto Takino; Kunihito Hoki
3RD INTERNATIONAL CONFERENCE ON APPLIED COMPUTING AND INFORMATION TECHNOLOGY (ACIT 2015) 2ND INTERNATIONAL CONFERENCE ON COMPUTATIONAL SCIENCE AND INTELLIGENCE (CSI 2015), IEEE, 97-102, 2015, Peer-reviwed, StarCraft: Brood War is one of most-popular classic real-time strategy (RTS) games and has attracted much interest in recent researches. The Artificial Intelligence (AI) in StarCraft has not been that advanced, so professional players have consistently been stronger than any AI. We have considered the build order problem in RTS games for this paper and propose a simple human-like build-order management model to pursue human intelligence, where matrix B represents the build order in the model. Our experiments using state-of-art AI showed that the model is capable of strengthening the AI and effective at countering a specific opponent's strategies. Our experiments also showed that switching B in the model in accordance with the information on the opponent's strategies from scouts is effective.
International conference proceedings, English

Analysis of Performance of Consultation Methods in Computer Chess
Kunehito Hoki; Seya Omori; Takeshi Ito
JOURNAL OF INFORMATION SCIENCE AND ENGINEERING, INST INFORMATION SCIENCE, 30, 3, 701-712, May 2014, Peer-reviwed, The performance of consultation methods, i.e., majority voting, the optimistic selection rule, and the pseudo-random number (PRN) ensemble method, are examined in computer chess using 2180 chess problems. Here, the optimistic selection rule selects a program that returns the highest search value, and the PRN ensemble consists of multiple individual copies of one base program, and each copy is diversified by adding random numbers to the evaluation function of the base program. We carried out empirical experiments by using state-of-the-art chess-program Crafty as the base program. We found that the percentage of correct answers increased from 55.6 to 57.1% using optimistic selection from the PEN ensemble. The experimental results indicated that the consultation methods allowed simple yet effective distributed computing in chess.
Scientific journal, English

Large-Scale Optimization for Evaluation Functions with Minimax Search
Kunihito Hoki; Tomoyuki Kaneko
JOURNAL OF ARTIFICIAL INTELLIGENCE RESEARCH, AI ACCESS FOUNDATION, 49, 527-568, 2014, Peer-reviwed, This paper presents a new method, Minimax Tree Optimization (MMTO), to learn a heuristic evaluation function of a practical alpha-beta search program. The evaluation function may be a linear or non-linear combination of weighted features, and the weights are the parameters to be optimized. To control the search results so that the move decisions agree with the game records of human experts, a well-modeled objective function to be minimized is designed. Moreover, a numerical iterative method is used to find local minima of the objective function, and more than forty million parameters are adjusted by using a small number of hyper parameters. This method was applied to shogi, a major variant of chess in which the evaluation function must handle a larger state space than in chess. Experimental results show that the large-scale optimization of the evaluation function improves the playing strength of shogi programs, and the new method performs significantly better than other methods. Implementation of the new method in our shogi program B o n a n z a made substantial contributions to the program's first-place finish in the 2013 World Computer Shogi Championship. Additionally, we present preliminary evidence of broader applicability of our method to other two-player games such as chess.
Scientific journal, English

Monte-Carlo simulation Adjusting
N. Araki; M. Muramatsu; K. Hoki; S. Takahashi
the Twenty-Eighth AAAI Conference on Artificial Intelligence, 3094-3095, 2014, Peer-reviwed
International conference proceedings, English

PARALLEL DOVETAILING AND ITS APPLICATION TO DEPTH-FIRST PROOF-NUMBER SEARCH
Kunihito Hoki; Tomoyuki Kaneko; Akihiro Kishimoto; Takeshi Ito
ICGA JOURNAL, IOS PRESS, 36, 1, 22-36, Mar. 2013, Peer-reviwed, Depth-first proof-number (df-pn) search is an effective sequential AND/OR tree search algorithm using the notion of proof and disproof numbers. Although df-pn has been parallelized in shared-memory environments thus far, parallelizing df-pn in distributed-memory environments still remains a challenge. This paper presents simple yet empirically effective parallel df-pn methods for distributed computing environments. Our methods are based on parallel dovetailing that has been successfully applied to a number of algorithms and they incur almost no communication overhead by independently performing df-pn search that exploits a different part of the search space in each processing node. More specifically, we present two methods of parallelizing an enhanced df-pn variant called df-pn(+) : the first is based on leveraging non-deterministic behaviors of a shared-memory parallel df-pn search algorithm (SPDDFPN+), and the second is based on randomly initializing proof and disproof numbers (RPDDFPN+). Experimental results using state-of-the-art solvers indicated that RPDDFPN+ achieves reasonable speedups in both tsume-shogi (checkmate problems in Japanese chess) and tsume-Go (life-and-death problems in Go), which have completely different characteristics. Moreover, parallel dovetailing occasionally yields superlinear speedups with a reasonably small number of base solvers and can even solve additional instances that the original solver is unable to solve. Our results also revealed that SPDDFPN(+)improves the efficiency of a high-performance tsume-shogi solver.
Scientific journal, English

A System-Design Outline of the Distributed-Shogi-System Akara 2010
Kunihito Hoki; Tomoyuki Kaneko; Daisaku Yokoyama; Takuya Obata; Hiroshi Yamashita; Yoshimasa Tsuruoka; Takeshi Ito
2013 14TH ACIS INTERNATIONAL CONFERENCE ON SOFTWARE ENGINEERING, ARTIFICIAL INTELLIGENCE, NETWORKING AND PARALLEL/DISTRIBUTED COMPUTING (SNPD 2013), IEEE, 0, 466-471, 2013, Peer-reviwed, This paper describes Akara 2010, the distributed shogi system that has defeated a professional shogi player in a public game for the first time in history. The system employs a novel design to build a high-performance computer shogi player for standard tournament conditions. The design enhances the performance of the entire system by means of distributed computing. To utilize a large number of computers, a majority-voting method using four existing programs is combined with a distributed-search method. Although the performance of the entire system could not be tested, the majority-voting component increased the winning percentage from 62% to 73%, and the distributed-search component increased it from 50% to 70% or more.
International conference proceedings, English

Distributed-Shogi-System Akara 2010 and its Demonstration
K. Hoki; T. Kaneko; D. Yokoyama; T. Obata; H. Yamashita; Y. Tsuruoka; T. Ito
International Journal of Computer & Information Science, International Association for Computer and Information Science (ACIS), 14, 55-63, 2013, Peer-reviwed
Scientific journal, English

Theoretical verification of nonthermal microwave effects on intramolecular reactions
Manabu Kanno; Kosuke Nakamura; Eri Kanai; Kunihito Hoki; Hirohiko Kono; Motohiko Tanaka
The Journal of Physical Chemistry, ACS Publications, 116, 2177-2183, Feb. 2012, Peer-reviwed
Scientific journal, English

Theoretical investigation of the structures and dynamics of crystalline molecular gyroscopes
Anant Babu Marahatta; Manabu Kanno; Kunihito Hoki; Wataru Setaka; Stephan Irle; Hirohiko Kono
Journal of Physical Chemistry C, ACS Publications, 116, 46, 24845-24854, 2012, Peer-reviwed, Recently, molecular rotor systems have been emerging as a promising candidate of functional nanoscale devices. A macroscopic gyroscope like molecule in a crystalline solid is particularly unique owing to its variable physicochemical properties. Setaka et al. have achieved the synthesis of a novel crystalline molecular gyroscope characterized by a closed topology with a phenylene rotator encased in three long siloxaalkane spokes [Setaka, W.
et al. Chem. Lett.2007, 36, 1076]. We theoretically investigated the underlying mechanism of its rotational dynamics by utilizing the self-consistent-charge density-functional-based tight-binding (DFTB) method for crystal structures. We first found that the DFTB semiquantitatively reproduced the unit cell molecular geometries of all three stable X-ray structures under the periodic boundary condition. From the potential energy surface calculations, the activation barrier for phenylene rotation was estimated to be about 1.2 kcal/mol, which is much lower than those of other, previously synthesized gyroscopic compounds. In comparison to 1,4-bis(trimethylsilyl)benzene of a similar crystal structure but of an open topology, the siloxaalkane frame in the crystalline molecular gyroscope under consideration effectively blocks strong intermolecular steric interactions experienced by the phenylene rotator. The molecular dynamics simulations based on the DFTB exemplified facile phenylene flipping between the stable structures, especially at high temperature. The present results demonstrate the remarkable ability of the DFTB method to predict the crystal structures and rotational dynamics of this type of crystalline molecular gyroscopes. © 2012 American Chemical Society.
Scientific journal, English

Performance Analysis of Consultation Methods in Computer Chess
Seiya Omori; Kunihito Hoki; Takeshi Ito
2012 CONFERENCE ON TECHNOLOGIES AND APPLICATIONS OF ARTIFICIAL INTELLIGENCE (TAAI), IEEE, 171-175, 2012, Peer-reviwed, The performance of consultation methods is examined in computer chess using two kinds of experiments. One is playing self-play games to observe the winning rates, and the other is solving a collection of chess problems to observe the percentages of the correct answers. It is shown that the winning rate of the optimistic selection rule with 4 base programs against the original one is 61%. Moreover, it is shown that the rate of correct answers with the nominal depth 8 increases from 59% to 70% using the optimistic selection rule with 16 base programs. These results indicate that consultation methods allow us simple yet effective distributed computing in chess.
International conference proceedings, English

Efficiency of three forward-pruning techniques in shogi: Futility pruning, null-move pruning, and Late Move Reduction (LMR)
Kunihito Hoki; Masakazu Muramatsu
Entertainment Computing, Elsevier B.V., 3, 3, 51-57, 2012, Peer-reviwed, The efficiency of three forward-pruning techniques, i.e., futility pruning, null-move pruning, and LMR, is analyzed in shogi, a Japanese chess variant. It is shown that the techniques with the α-β pruning reduce the effective branching factor of shogi endgames to 2.8 without sacrificing much accuracy of the search results. Because the average number of the raw branching factor in shogi is around 80, the pruning techniques reduce the search space more effectively than in chess. © 2011 International Federation for Information Processing.
Scientific journal, English

Analysis of evaluation-function learning by comparison of sibling nodes
Tomoyuki Kaneko; Kunihito Hoki
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), 7168, 158-169, 2012, Peer-reviwed, This paper discusses gradients of search values with a parameter vector θ in an evaluation function. Recent learning methods for evaluation functions in computer shogi are based on minimization of an objective function with search results. The gradients of the evaluation function at the leaf position of a principal variation (PV) are used to make an easy substitution of the gradients of the search result. By analyzing the variations of the min-max value, we show (1) when the min-max value is partially differentiable and (2) how the substitution may introduce errors. Experiments on a shogi program with about a million parameters show how frequently such errors occur, as well as how effective the substitutions for parameter tuning are in practice. © 2012 Springer-Verlag.
International conference proceedings, English

The global landscape of objective functions for the optimization of shogi piece values with a game-tree search
Kunihito Hoki; Tomoyuki Kaneko
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), 7168, 184-195, 2012, Peer-reviwed, The landscape of an objective function for supervised learning of evaluation functions is numerically investigated for a limited number of feature variables. Despite the importance of such learning methods, the properties of the objective function are still not well known because of its complicated dependence on millions of tree-search values. This paper shows that the objective function has multiple local minima and the global minimum point indicates reasonable feature values. Moreover, the function is continuous with a practically computable numerical accuracy. However, the function has non-partially differentiable points on the critical boundaries. It is shown that an existing iterative method is able to minimize the functions from random initial values with great stability, but it has the possibility to end up with a non-reasonable local minimum point if the initial random values are far from the desired values. Furthermore, the obtained minimum points are shown to form a funnel structure. © 2012 Springer-Verlag.
International conference proceedings, English

Experimental and theoretical investigations of isomerization reactions of ionized acetone and its dimer
Yoshiyuki Matsuda; Kunihito Hoki; Satoshi Maeda; Ken-ichi Hanaue; Keisuke Ohta; Keiji Morokuma; Naohiko Mikami; Asuka Fujii
PHYSICAL CHEMISTRY CHEMICAL PHYSICS, ROYAL SOC CHEMISTRY, 14, 2, 712-719, 2012, Peer-reviwed, Ionization dynamics of acetone and its dimer in supersonic jets is investigated by a combination of experimental and theoretical techniques, both of which have recently been developed. In experiments, the neutral and the cationic structures are explored by infrared predissociation spectroscopy with the vacuum-ultraviolet photoionization detection schemes. Reaction paths following the one-photon ionization of the acetone monomer and its dimer have been studied by the joint use of several theoretical methods including the ab initio molecular dynamics, the global reaction route mapping, the intrinsic reaction coordinate, and the artificial force induced reaction calculations. Upon one-photon ionization, the dimer isomerizes to the H-bonded form, in which the enol cation of acetone is bound to the neutral molecule, while this enolization is energetically forbidden in the acetone monomer. The enolization of the dimer cation occurs through a two-step proton-transfer from the methyl group of the ionized moiety, and is catalyzed by the neutral moiety within the dimer cation.
Scientific journal, English

Excitation of Biomolecules by Coherent vs. Incoherent Light: Model Rhodopsin Photoisomerization
Kunihito Hoki; Paul Brumer
Procedia Chemistry, Elsevier BV, 3, 1, 122-131, 2011, Peer-reviwed
Scientific journal

Consultation Algorithm for Computer Shogi: Move Decisions by Majority
Takuya Obata; Takuya Sugiyama; Kunituto Hoki; Takeshi Ito
COMPUTERS AND GAMES, SPRINGER-VERLAG BERLIN, 6515, 156-+, 2011, Peer-reviwed, A new algorithm that runs on a computer with interconnected processors has been designed for Shogi. The algorithm adopts consultation between many individual players. A method that can create multiple players from one program is presented. Applying a simple rule to select a decision on a single move, the consultation algorithm improves the performance of computer Shogi engines. It is also demonstrated that a council system consisting of three well-known Shogi programs: YSS, GPS, and BONANZA plays better games than any of the three programs individually.
International conference proceedings, English

Optimistic Selection Rule Better Than Majority Voting System
Takuya Sugiyama; Takuya Obata; Kunihito Hoki; Takeshi Ito
COMPUTERS AND GAMES, SPRINGER-VERLAG BERLIN, 6515, 166-+, 2011, Peer-reviwed, A recently proposed ensemble approach to game-tree search has attracted a great deal of attention. The ensemble system consists of M computer players, where each player uses a different series of pseudo-random numbers. A combination of multiple players under the majority voting system would improve the performance of a Shogi-playing computer. We present a new strategy of move selection based on the search values of a number of players. The move decision is made by selecting one player from all M players. Each move is selected by referring to the evaluation value of the tree search of each player. The performance and mechanism of the strategy are examined. We show that the optimistic selection rule, which selects the player that yields the highest evaluation value, outperforms the majority voting system. By grouping 16 or more computer players straightforwardly, the winning rates of the strongest Shogi programs increase from 50 to 60% or even higher.
International conference proceedings, English

将棋における合議アルゴリズム―局面評価値に基づいた指し手の選択
杉山卓弥; 小幡拓弥; 斉藤博昭; 保木邦仁; 伊藤毅志
情報処理学会論文誌, 情報処理学会, 51, 11, 2048-2054, 2010, Peer-reviwed
Scientific journal, Japanese

将棋における合議アルゴリズム―多数決による手の選択―
Takeshi Ito; Takuya Obata; Takuya Sugiyama; Kunihito Hoki
情報処理学会論文誌, 52, 11, 3030-3037, 2010, Peer-reviwed
Scientific journal, Japanese

Quantum dynamics of light-driven chiral molecular motors
Masahiro Yamaki; Shin-ichiro Nakayama; Kunihito Hoki; Hirohiko Kono; Yuichi Fujimura
PHYSICAL CHEMISTRY CHEMICAL PHYSICS, ROYAL SOC CHEMISTRY, 11, 11, 1662-1678, 2009, Peer-reviwed, The results of theoretical studies on quantum dynamics of light-driven molecular motors with internal rotation are presented. Characteristic features of chiral motors driven by a non-helical, linearly polarized electric field of light are explained on the basis of symmetry argument. The rotational potential of the chiral motor is characterized by a ratchet form. The asymmetric potential determines the directional motion: the rotational direction is toward the gentle slope of the asymmetric potential. This direction is called the intuitive direction. To confirm the unidirectional rotational motion, results of quantum dynamical calculations of randomly-oriented molecular motors are presented. A theoretical design of the smallest light-driven molecular machine is presented. The smallest chiral molecular machine has an optically driven engine and a running propeller on its body. The mechanisms of transmission of driving forces from the engine to the propeller are elucidated by using a quantum dynamical treatment. The results provide a principle for control of optically-driven molecular bevel gears. Temperature effects are discussed using the density operator formalism. An effective method for ultrafast control of rotational motions in any desired direction is presented with the help of a quantum control theory. In this method, visible or UV light pulses are applied to drive the motor via an electronic excited state. A method for driving a large molecular motor consisting of an aromatic hydrocarbon is presented. The molecular motor is operated by interactions between the induced dipole of the molecular motor and the electric field of light pulses.
Scientific journal, English

Dissipation effects on laser control of cis/trans isomerization
Kunihito Hoki; Paul Brumer
CHEMICAL PHYSICS LETTERS, ELSEVIER SCIENCE BV, 468, 1-3, 23-27, Jan. 2009, Peer-reviwed, Photoisomerization in a dissipative environment is examined using optimal control within the secular Redfield approximation. With decoherence assumed rapid compared to relaxation, we find that as the environment-induced dissipation increases the optimal pulse moves towards earlier pulse times ( to take advantage of relaxation effects), shorter duration, increased power, with lower resultant target cross sections. The control mechanism is seen to change from pump-dump to pump-dump-plus-environmentally-induced-relaxation as the coupling to the environment increases. (C) 2008 Elsevier B. V. All rights reserved.
Scientific journal, English

Quantum control of a chiral molecular motor driven by femtosecond laser pulses: Mechanisms of regular and reverse rotations
M. Yamaki; K. Hoki; H. Kono; Y. Fujimura
CHEMICAL PHYSICS, ELSEVIER SCIENCE BV, 347, 1-3, 272-278, May 2008, Peer-reviwed, Rotational mechanisms of a chiral molecular motor driven by femtosecond laser pulses were investigated on the basis of results of a quantum control simulation. A chiral molecule, (R)-2-methyl-cyclopenta-2,4-dienecarboaldehyde, was treated as a molecular motor within a one-dimensional model. It was assumed that the motor is fixed on a surface and driven in the low temperature limit. Electric fields of femtosecond laser pulses driving both regular rotation of the molecular motor with a plus angular momentum and reverse rotation with a minus one were designed by using a global control method. The mechanism of the regular rotation is similar to that obtained by a conventional pump-dump pulse method: the direction of rotation is the same as that of the initial wave packet propagation on the potential surface of the first singlet (n pi*) excited state S-1. A new control mechanism has been proposed for the reverse rotation that cannot be driven by a simple pump-dump pulse method. In this mechanism, a coherent Stokes pulse creates a wave packet localized on the ground state potential surface in the right hand side. The wave packet has a negative angular momentum to drive reverse rotation at an early time. (C) 2007 Elsevier B.V. All rights reserved.
Scientific journal, English

Quantum optimal control of electron ring currents in chiral aromatic molecules
Manabu Kanno; Kunihito Hoki; Hirohiko Kono; Yuichi Fujimura
JOURNAL OF CHEMICAL PHYSICS, AMER INST PHYSICS, 127, 20, 204314-1-204314-9, Nov. 2007, Peer-reviwed, We report the results of optimal control simulations of pi-electron rotation (ring current) in a six-membered chiral aromatic molecule, 2,5-dichloro[n](3,6)pyrazinophane (DCP), attached at a surface and excited by a linearly polarized UV laser. DCP has a pair of optically allowed, quasidegenerate pi-electronic excited states. The laser pulse to generate an approximate angular momentum eigenstate consisting of the quasidegenerate states was designed using the global optimal control theory. For both counterclockwise and clockwise pi-electron rotations, the calculated objective functional and target yield as a function of the angle of the photon polarization vector show two maxima and two minima. The origin of the two minima is coherent excitation to only one of the quasidegenerate states. The two maxima arise from creation of a superposition of the quasidegenerate states. The optimal control pulse at the maxima is a two-color laser field resonant with the quasidegenerate states. The electric field of the optimal control pulse consists of two parts: a slowly oscillating part with phase phi(env) and a rapidly oscillating one. The phase phi(env) is a crucial parameter for determination of the rotation direction of pi electrons at the end of control. The results of the optimal control simulations suggest that pi-electron rotation can be controlled by applying a two-color laser field with adjusted phases. (c) 2007 American Institute of Physics.
Scientific journal, English

Theoretical Design of an Aromatic Hydrocarbon Rotor Driven by a Circularly Polarized Electric Field
Masahiro Yamaki; Kunihito Hoki; Takato Teranishi; Wilfredo Credo Chung; Fabio Pichierri; Hirohiko Kono; Yuichi Fujimura
Journal of Physical Chemistry A, American Chemical Society, 111, 9374-9378, Aug. 2007, Peer-reviwed
Scientific journal, English

Calculation of the cross section of epithermal neutron scattering from water by a time-dependent wavepacket approach
Seiichi Tanabe; Kunihito Hoki; Hirohiko Kono; Yuichi Fujimura
CHEMICAL PHYSICS LETTERS, ELSEVIER SCIENCE BV, 437, 4-6, 267-271, Apr. 2007, Peer-reviwed, An expression for an epithermal neutron scattering cross section from a water cluster is derived by using a time-dependent wavepacket approach. Both the translational motion of the center of mass of the cluster and the vibrational modes of a proton in the cluster are taken into account. Incident energy dependence on the neutron proton scattering cross sections calculated for a simplified cluster model agrees with experimental results of neutron scattering from water. Time correlation functions of momentum transfer clearly show that both the translational and vibrational motions significantly contribute to the scattering cross section in the low energy regime. (c) 2007 Elsevier B.V. All rights, reserved.
Scientific journal, English

An analysis of two liquid-state adaptive feedback experiments
Paul Brumer; Kunihito Hoki; Michael Spanner
ISRAEL JOURNAL OF CHEMISTRY, WILEY-BLACKWELL, 47, 1, 111-114, 2007, Peer-reviwed, Adaptive feedback experiments, which provide an engineering oriented approach to optimizing a desired target in a molecular process, have been extensively developed over the past decade. Efforts to extract the control mechanisms that underlie the observed control have, however, only recently begun in earnest. We describe the results of the analyses of two liquid state adaptive feedback experiments, showing that the resultant control takes little advantage of coherent quantum effects.
Scientific journal, English

Mechanisms in Adaptive Feedback Control: Photoisomerization in a Liquid
Kunihito Hoki; Paul Brumer
Physical Review Letters, American Physical Society (APS), 95, 16, 12 Oct. 2005, Peer-reviwed
Scientific journal

Quantum control of unidirectional rotations of a chiral molecular motor
M. Yamaki; K. Hoki; Y. Ohtsuki; H. Kono; Y. Fujimura
Phys. Chem. Chem. Phys., 7, 1900-1904, Apr. 2005, Peer-reviwed
English

Quantum control of a chiral molecular motor driven by laser pulses
M. Yamaki; K. Hoki; Y. Ohtsuki; H. Kono; Y. Fujimura
J. Am. Chem. Soc., 127, 7300-7301, Apr. 2005, Peer-reviwed
English

Quantum control of a chiral molecular motor driven by linearly polarized laser pulses
M. Yamaki; K. Hoki; Y. Ohtsuki; H. Kono; Y. Fujimura
in “Ultrafast Phenomena XIV”, 508-510, 2005, Peer-reviwed
English

Sequential pump-dump control of photoisomerization competing with photodissociation of optical isomers
Kunihito Hoki; Leticia González; Mohamed F. Shibl; Yuichi Fujimura
Journal of Physical Chemistry A, 108, 31, 6455-6463, 05 Aug. 2004, Peer-reviwed, A sequential pump-dump control scenario for photoisomerization competing with direct photodissociation of a chiral molecule is derived by using a global optimization method. As an example, H2POSD with axial chirality is considered. The first excited electronic state with n-σ* electronic configuration is characterized by a repulsive potential that leads to P-S bond breaking. In this scenario, a first pump-dump sequence creates a nuclear wave packet having enough kinetic energy to cross the potential barrier in the ground state. This wave packet moves from one potential well corresponding to one of the optical isomers to the other in the electronic ground state. This transfer minimizes effects of the competing photodissociation. The second pump-dump sequence converts the wave packet into one having a small amount of kinetic energy via excitation to the first excited state. A stimulated Raman adiabatic passage (STIRAP) is applied to the same system for comparison. The yield of the photoisomerization obtained by using the STIRAP method is less than 0.1%, while that of using the sequential pump-dump pulse method is about 25%.
Scientific journal, English

Mechanism of unidirectional motions of chiral molecular motors driven by linearly polarized pulses
Kunihito Hoki; Masahiro Yamaki; Shiro Koseki; Yuichi Fujimura
Journal of Chemical Physics, 119, 23, 12393-12398, 15 Dec. 2003, Peer-reviwed, The mechanism of unidirectional motion of randomly oriented chiral molecular motors driven by linearly polarized laser pulses was clarified by analyzing the contour plots of the instantaneous angular momentum averaged over their orientations. The instantaneous angular momentum was estimated by using a quantum master equation approach and a classical approach of a trajectory calculation.
Scientific journal, English

Chiral molecular motors driven by a nonhelical laser pulse
Kunihito Hoki; Masahiro Yamaki; Yuichi Fujimura
Angewandte Chemie - International Edition, 42, 26, 2976-2978, 07 Jul. 2003, Peer-reviwed, A turn for the better? The unidirectional rotation of chiral molecular motors, driven by a nonhelical laser pulse, can be explained by both quantum and classical mechanical calculations. The aldehyde group of the molecule is the rotor engine. The motor dynamics consists of three regimes: ignition, acceleration, and rotation. The results serve as a guiding principle for the control of molecular.
Scientific journal, English

Molecular motors driven by laser pulses: Role of molecular chirality and photon helicity
Kunihito Hoki; Masahiro Yamaki; Shiro Koseki; Yuichi Fujimura
Journal of Chemical Physics, 118, 2, 497-504, 08 Jan. 2003, Peer-reviwed, A study was conducted on the mechanisms of the unidirectional motion of molecular motors driven by a linearly or circularly polarized laser field within the dipole interaction approximation. As such, pure chiral molecules with internal rotation was considered as its engine. An expression for estimation of the instantaneous angular momentum of the internal rotation of randomly oriented molecular motors, which expresses the rotational wave packet at time t was derived. It was shown that the ensemble-averaged angular momentum of pure enantiomers reverses its sign by substitution by the other pure enantiomers regardless of the helicity of laser fields.
Scientific journal, English

Quantum control of molecular handedness in a randomly oriented racemic mixture using three polarization components of electric fields
Kunihito Hoki; Leticia González; Yuichi Fujimura
Journal of Chemical Physics, 116, 20, 8799-8802, 22 May 2002, Peer-reviwed, The quantum control of molecular handedness in a randomly oriented racemic mixture using three polarization components of electric fields was discussed. One of the three polarization components was used for orientation and the other two were used for controlling the chirality. The fundamental principle is the adjustment of the polarization direction of pulses to the dipole moment interactions. The effectiveness of the method was demonstrated by numerical simulations on the enantiomeric enrichment of the axial chiral H2POSH molecule.
Scientific journal, English

Control of molecular handedness using pump-dump laser pulses
Kunihito Hoki; Leticia González; Yuichi Fujimura
Journal of Chemical Physics, 116, 6, 2433-2438, 08 Feb. 2002, Peer-reviwed, A method was proposed to control molecular handedness of preoriented enantiomers using linearly polarized femtosecond pump-dump laser pulses. The direction of electronic transition moment vector was used to determine the molecular handedness of the enantiomers. The density functional theory was used to obtain electronic ground and singlet excited state potentials and transition dipole moments of H2POSH racemic mixture. Analysis suggested that linearly polarized lasers efficiently controlled the molecular handedness of preoriented molecules.
Scientific journal, English

Quantum Control of NaI Predissociation in Subpicosecond and Several-Picosecond Time Regimes
K. Hoki; Y. Ohtsuki; H. Kono; Y. Fujimura
Journal of Physical Chemistry A, 103, 32, 6301-6308, Jul. 1999, Peer-reviwed
Scientific journal, English

Quantum control of the photodissociation of sodium iodide
K. Hoki; Y. Ohtsuki; H. Kono; Y. Fujimura; S. Koseki
Bulletin of the Chemical Sociey of Japan, 72, 12, 2665-2671, 1999, Peer-reviwed
Scientific journal, English

AobaZero の高速化と現在の状況
山下宏; 保木邦仁; 小林祐樹
コンピュータ将棋協会, Mar. 2021, コンピュータ将棋協会誌, 32, 72-86, Japanese, Introduction other

深層畳み込みニューラルネットワークによる 自動車用オートマチックトランスミッションの制御に現れる時系列データの2値分類
井手貴範; 川上雄史; 保木邦仁; 村松正和
日本応用数理学会, Sep. 2020, 応用数理 インダストリアルマテリアルズ, 30, 3, 120-124, Japanese, Introduction scientific journal

AobaZero の紹介と観戦記
山下宏; 保木邦仁; 小林祐樹
Mar. 2020, コンピュータ将棋協会誌, 31, 38-44, Japanese, Introduction other

"Machine Learning of Positional Evaluation, Brute-Force Search, and Consultation Method in Computer Shogi"
Kunihito Hoki
Sep. 2012, Journal of the Japan Society for Management Information, 21, 2, 97-100, Japanese, Introduction other

Unexpected Common Ground between Chemistry and Game Informatics
Kunihito Hoki
This article reviews recent advances in computer games. It is explained that methods famous in computational chemistry have enhanced the performance of Go and shogi programs. A history of the grand challenge in computer science that aims to defeat human experts in games is briefly summarized., Japan Society for Molecular Science, 2012, Molecular Science, 6, 1, A0050, Japanese, Peer-reviwed, Introduction other, 1881-8404, 130002126146

コンピュータ将棋が採用している局面評価の機械学習・力づく探索・合議法
保木邦仁
経営情報学会誌編集事務局, 2012, 経営情報学会誌, 21, 2, 97-100, Japanese, Introduction other, 0918-7324, 40019453877, AN10551813

Effective Branching Factors of Bonanza and Crafty (Frontiers of Computer Shogi (1))
Kunihito Hoki
人工知能学会, May 2011, Journal of Japanese Society for Artificial Intelligence, 26, 3, 295–300-300, Japanese, Introduction other, 0912-8085, 110008662159, AN10067140

URL

Way to Victory of AKARA2010 : 2. An Outline of the System Design and Operation of Akara 2010
Kunihito Hoki; Tomoyuki Kaneko; Daisaku Yokoyama; Takuya Obata; Hiroshi Yamashita
Feb. 2011, IPSJ Magazine, 52, 2, 162–168, Japanese, Introduction other

Way to Victory of AKARA2010 : 4. Ichiyo Shimizu vs Akara 2010 : Reviewing the Computers' Thinking Processes
Yoshimasa Tsuruoka; Tomoyuki Kaneko; Hiroshi Yamashita; Kunihito Hoki
情報処理学会, Feb. 2011, IPSJ Magazine, 52, 2, 175-180, Japanese, Introduction other, 0447-8053, 110008507309, AN00116625

Landscape for optimization problem of minimax search
保木 邦仁; 金子 知適
将棋において局面評価関数の機械学習が成功を収めている。この学習法では、局面評価関数の特徴ベクトルvを調整してminimax探索を棋譜データに沿って最適化する。本研究では、この最適化に用いられる関数J(v)の形状を、連続性、偏微分不可能性、極小点をどの程度持つかという点に焦点を絞り解析し、結果を報告する。In Shogi, machine learning of the positional evaluation function has proved successful. The learning method adjusts the feature vector v of the positional evaluation function by solving an optimization problem of minimax search in accordance with professional game scores. In this paper, we analyze properties of the function J(v) to be optimized in terms of continuity, partial differentiability, and number of local minima., 12 Nov. 2010, ゲームプログラミングワークショップ2010論文集, 2010, 12, 67-70, Japanese, 170000063551, AA12496601

全遷移状態の自動探索アルゴリズム
2009, 化学同人 化学, 64, 70-71, Japanese, Introduction scientific journal

Optical control of chiral molecular motors
Kunihito Hoki; Masahiro Yamaki; Yuichi Fujimura
Results of theoretical treatments of optical control of chiral molecular motors driven by linearly polarized laser pulses are presented. Both quantum and classical simulations of time-dependent quantum mechanical expectation values were performed to identify the rotational direction of real molecular motors. The rotational direction was found to be toward the gentle slope of the asymmetric potential energy surface of a chiral molecule of interest, which is the intuitive direction of rotation. The mechanism of its unidirectional motion is a non-resonant multi-photon forced-rotation induced by linearly polarized intense laser pulses. Relaxation effects of randomly oriented molecular motors were investigated using the Lindblad-type quantum master equation. Quantum control of unidirectional rotation of a chiral molecular motor is presented. Counter-intuitive rotation as well as intuitive rotation is generated by using a quantum control theory. Time- and frequency-resolved spectra of the electric field of the optimal laser were evaluated to analyze the origin of both intuitive and counter-intuitive rotations. A femtosecond pump-dump control method via an electronic excited state is shown to be one of the effective methods for avoiding effects of couplings between motors and solvents. © 2008 Springer-Verlag., 2008, Springer Series in Chemical Physics, 89, 93-112, English, Book review, 0172-6218, 59249089441

Control of Molecular Chirality by Lasers
2008, Advances in Multi-Photon Processes and Spectroscopy, 18, 245-290, Japanese, Introduction scientific journal

ーザーによるキラル分子の回転方向制御のシミュレーション：光駆動キラル分子モーター
2008, 化学同人 最新分子マシン, ,128-131, Japanese, Introduction commerce magazine

コンピュータ将棋における全幅探索とfutility pruningの応用
Kunihito Hoki
2006, IPSJ Magazine, 47, 884–889, Japanese, Introduction other

ゲーム情報学概論－ゲームが切り拓く人工知能－
伊藤毅志; 保木邦仁; 三宅一郎
Textbook, Japanese, Joint work, 第II部ゲーム情報学のアルゴリズム, コロナ社, 2018

人工知能学大辞典
保木邦仁
Dictionary or encycropedia, Japanese, Contributor, 章：ゲーム、見出し：Bonanza, 人工知能学会, Jul. 2017

文殊の誕生，あから2010の人間への挑戦，人間に勝つコンピュータ将棋の作り方【あから2010を生み出したアイデアと工夫の軌跡】
Kunihito Hoki
Japanese, Joint work, 数の暴力で人間に挑戦！ Bonanzaの誕生, Gijyutsu Hyoronsya, Nov. 2012

数の暴力で人間に挑戦！ Bonanza の誕生，人間に勝つコンピュータ将棋の作り方【あから2010を生み出したアイデアと工夫の軌跡】
Kunihito Hoki
Japanese, Joint work, 数の暴力で人間に挑戦！ Bonanzaの誕生, Gijyutsu Hyoronsya, Nov. 2012

Advances in Computer Shogi
Kunihito Hoki
Japanese, Joint work, Bonanza 4.1.3 ソースコードの解説, kyoritsu publishing, May 2012

ボナンザVS勝負脳―最強将棋ソフトは人間を超えるか
保木邦仁; 渡辺明
Japanese, Joint work, 角川書店, Aug. 2007

着手の不可逆的性質を利用したDFPN探索によるYonmoqueの求解
中川憲; 荒井光; 保木邦仁; 村松正和
Oral presentation, Japanese, 情報処理学会ゲーム情報学(GI),研究報告 2013-GI-29 No.6, Domestic conference
2013

コンピュータ大貧民に対する差分学習法の応用
小沼啓; 本多武尊; 保木邦仁; 西野哲朗
Others, Japanese, 情報処理学会ゲーム情報学(GI),研究報告 2012-GI-27 No. 1
2012

チェスプログラムを用いた合議アルゴリズムの効果の検証
大森誠也; 保木邦仁; 伊藤毅志
Others, Japanese, 情報処理学会ゲーム情報学(GI),研究報告, 26, pp. 17, 2011
2011

Special Lecture on Informatics and Engineering A （Learn to Play Games Called Mind Sports ）
The University of Electro-Communications

西東京三大学英語化科目 学域特別講義A（マインドスポーツゲーム を学ぶ）
電気通信大学

Graduate Technical English
The University of Electro-Communications

大学院技術英語
電気通信大学

Advanced Internship
The University of Electro-Communications

大学院インターンシップ
電気通信大学

Internship
The University of Electro-Communications

インターンシップ（Ⅰ類）
電気通信大学

AI・セキュリティー 人材育成プログラム ゲーム人工知能概論
電気通信大学 社会人履修証明プログラム

AI・セキュリティー 人材育成プログラム ゲーム人工知能概論
電気通信大学 社会人履修証明プログラム

MICS Science Laboratory II A, B
The University of Electro-Communications

MICS 実験第二 A, B
電気通信大学

game informatics
The University of Electro-Communications

Special Lecture on Informatics and Engineering A(Learn to Play Games Called Mind Sports)
The University of Electro-Communications

学域特別講義A（マインドスポーツゲームを学ぶ）
電気通信大学

ゲームの数理
The University of Electro-Communications

総合コミュニケーション科学模擬授業
The University of Electro-Communications

イノベイティブ総合コミュニケーションデザイン１
The University of Electro-Communications

イノベイティブ総合コミュニケーションデザイン１
電気通信大学

ゲームの数理
電気通信大学

世界強要プログラム：現代社会の諸問題において顕在化する科学技術
東京外語大学

総合コミュニケーション科学模擬授業
電気通信大学

世界強要プログラム：現代社会の諸問題において顕在化する科学技術
東京外語大学

世界強要プログラム：現代社会の諸問題において顕在化する科学技術
東京外語大学

情報通信演習2
The University of Electro-Communications

情報通信演習2
電気通信大学

情報数理工学実験第二
The University of Electro-Communications

情報数理工学実験第二
電気通信大学

ゲーム情報学
The University of Electro-Communications

ゲーム情報学
電気通信大学

応用数学A
The University of Electro-Communications

応用数学A
電気通信大学

The Chemical Society of Japan

Japan Society of Molecular Science

Information Processing Society of Japan

The Japanese Society for Artificial Intelligence

将棋人工知能に関するディープマインド社の大規模実験とその知識獲得過程の検証
保木 邦仁; 伊藤 毅志
日本学術振興会, 科学研究費助成事業, 電気通信大学, 基盤研究(C), Silver らは 2019 年、囲碁・将棋などの知識を自己対局形式により獲得する人工知能プレイヤの強化学習アルゴリズム AlphaZero とその実験結果をサイエンス誌で発表した。本研究では、この先行研究の将棋における大規模実験の追試を市場に出回っているハードウエアを用いて行い、強化学習過程や生成された人工知能の性能を観測し、AlphaZero が将棋知識を獲得する過程を分析する。 申請者は追実験を行うため、グラフィックス・プロセッシング・ユニット (GPU) を活用した自己対局生成の計算効率を追求した。NVIDIA 社の一個15万円程度の GPU (GeForce RTX 2080) を用いて、日産１万局程度の効率を達成した。現在申請者が参加している追実験のプロジェクトでは、市販ハードウエアのみを用いて構築された追実装は、AlphaZero の将棋の実験に準じた性能を獲得している (山下宏、保木邦仁、小林祐樹、AobaZero の高速化と現在の状況、コンピュータ将棋協会誌、vol. 32, pp. 72-86, 2021)。 申請者は現在、計算機実験を行う環境整備に取り組み、性能評価を行うための実装について検討中である。, 20K12120
01 Apr. 2020 - 31 Mar. 2023

人智を超えるゲームAIを利用した知の拡張
伊藤 毅志; 保木 邦仁; 山本 雅人; 松原 仁; 池田 心
日本学術振興会, 科学研究費助成事業, 電気通信大学, 基盤研究(B), 本研究の目標は、ゲームAIと人間の思考の違いを理解して、人智を超えたゲームAIの思考を人間に理解しやすい形に表示して、人間の知を拡張する手法を提案し、その効果について認知的評価を試みることである。 その目標を達成するために、ゲームAIの強化と思考の可視化について、研究を進めてきた。将棋の分野では、人間を超える将棋AIの出現が人間に与える影響に関する研究を進めており、そのテストケースとして、小路盤の将棋としての５五将棋を題材として、人間を遥かに凌駕するレベルのAIを実現した。このAIを用いることで、将棋の亜種のゲームに関する様々な知見を深めることが出来た。また、人間を超える将棋AIの出現前後のトッププロ棋士の棋譜と将棋AIとの手の一致率を調べる研究を行って、何らかの影響を受けている可能性があることが示唆された。 囲碁の分野では、ポリシーネットワークの可視化技術について、研究を行った。人間の理解の困難な深層学習の可視化技術は、現在様々な形で求められている。多人数不完全情報ゲームである麻雀に関してもAIプレイヤの強化を図っている。 カーリングの分野では、実環境のデータを組み込むためのシミュレーターの基礎研究を進めており、高度な不確定情報を含むゲームにおける実環境データの再現を試みている。人狼を対象とした分野では、人間特有の意思決定過程に関するモデルの提案を行った。人狼AIへの応用を考えるとともに、高度に複雑なコミュニケーションを必要とする人狼における知の拡張について考察を深めている。, 18H03347
01 Apr. 2018 - 31 Mar. 2023

Applying heuristics to complex games close to real-world competition
Hoki Kunihito
Japan Society for the Promotion of Science, Grants-in-Aid for Scientific Research, The University of Electro-Communications, Grant-in-Aid for Scientific Research (C), Aiming to develop competitive artificial intelligence by achieving the scale-up of existing heuristics in games that are close to competition in the real world, where there are many options for actions, and situations and state transitions can only be partially observed. This research was carried out by using Go, Mahjong, a card game, chess, digital curling, etc., provided a new algorithm to realize the aim, and was published in several papers., 16K00503
01 Apr. 2016 - 31 Mar. 2020

Development of ensemble game tree search method
Hoki Kunihito
Japan Society for the Promotion of Science, Grants-in-Aid for Scientific Research, The University of Electro-Communications, Grant-in-Aid for Young Scientists (B), We analyzed ensemble methods to avoid decision error of artificial intelligence. Specifically, we took mind sports games such as chess as research targets, and aimed at elimination of blunders by composing an ensemble of artificial intelligence. The major outcomes of this research were published as: (i) K. Hoki, T. Kaneko, A. Kishimoto, T. Ito, Parallel Dovetailing and Its Application to Depth-First Proof Number Search, ICGA Journal, 36, 22-36, 2013. (ii) K. Hoki, S. Omori, T. Ito, Analysis of Performance of Consultation Methods in Computer Chess, Journal of Information Science and Engineering, 30, 701-712, 2014., 25730212
01 Apr. 2013 - 31 Mar. 2016

Research on the game AI that makes human-lile mistakes
Takeshi Ito; HOKI KUNIHITO; NISHINO TETSURO; MUNEKATA NAGISA; KATAYOSE HARUHIRO; IKEDA KOKOLO
Japan Society for the Promotion of Science, Grants-in-Aid for Scientific Research, The University of Electro-Communications, Grant-in-Aid for Scientific Research (B), In this research, we focused on mistakes of human. We got the following research results by aiming at making game AI that makes human-like mistakes. 1) Classification by focusing on the cause of a mistake in human playing games.2) Proposal of a game AI with the model in consideration of human biological restrictions.3) Proposal and estimation of a game AI that adjust the strength in a game automatically and direct to a good game.4) Proposal of a game AI that realizes a human-like play by giving the "flow". These research results bring new evaluation other than the directivity of "strength" to game AI and serve as new indicator to generate the various game AI., 25280130
01 Apr. 2013 - 31 Mar. 2016

Simulating brain functions based on computational learning theory
NISHINO Tetsuro; TANAKA Shigeru; YAMAZAKI Tadashi; HOKI Kunihito
Japan Society for the Promotion of Science, Grants-in-Aid for Scientific Research, The University of Electro-Communications, Grant-in-Aid for Scientific Research (B), In this study, we construct various models of brain functions, such as motion, thinking and memory based on computational learning theory, and analyze their behavior in the following fashion: (1) we construct a state transition model for the motion and simulate the robot motion using this model, (2) we construct a neural network model for the memory and analyze its behavior, and (3) we design an efficient algorithm for card game playing using Monte Carlo Simulation. Finally we construct a unified methodology for simulating brain functions based on the above results., 23300055
01 Apr. 2011 - 31 Mar. 2014

Dynamics of molecules of multiple electronic properties in laser fields and their control
KONO Hirohiko; OHTSUKI Yukiyoshi; HOKI Kunihito
Japan Society for the Promotion of Science, Grants-in-Aid for Scientific Research, Tohoku University, Grant-in-Aid for Scientific Research on Priority Areas, We investigated the dynamics of molecules interacting with near-infrared intense laser pulses by using the time-dependent adiabatic state approach developed. The results for C60 indicate that the mode selectivity of vibrational excitation can be achieved by a pulse train, i.e., by changing the intervals between pulses. Time-dependent multiconfiguration theory is also developed to investigate the multielectron (ionization) dynamics of molecules. We have demonstrated that nuclear vibration is nonadiabatically coupled to the ring current of π electrons., 18066002
2006 - 2009

Theoretical design and analysis of nanomachines by using molecular dynamics control technique
HOKI Kunihito
Japan Society for the Promotion of Science, Grants-in-Aid for Scientific Research, Tohoku University, Grant-in-Aid for Young Scientists (Start-up), 研究代表者がこれまでに研究してきた分子ダイナミクス制御・レーザー制御理論に基づいた, ナノマシンの理論設計・実験結果の解析・機能分子の運動の観測手段の提案を目指して研究を進めた. 特に, 分子モーターの人工合成に必要と思われる基礎理論の構築を目指した研究を行った. また, 第一原理分子動力学法によりモーター分子の運動を評価して回転運動の伝達や散逸の機構を明らかにした., 19850001
2007 - 2008

Theoretical study on quantum control of photo-induced electron and nuclear dynamics of chiral molecules and generation of their functionality
FUJIMURA Yuichi; KONO Hirohiko; HOKI Kunihito
Japan Society for the Promotion of Science, Grants-in-Aid for Scientific Research, Tohoku University, Grant-in-Aid for Scientific Research (B), Recently considerable attention has been directed to control of ultra-fast electron and nuclear dynamics in photo-excited molecules. Especially, electron and unclear dynamics of chiral molecules are fascinating targets for generation of functionality such as molecular motors, switching and electron currents since these are induced by asymmetric electronic and nuclear potentials. In this project, we paid our attention to two subjects: (a) theoretical design of femto-second unidirectional rotation of molecular motor by using a quantum control method; (b) theoretical proposal of generation of pi-electron currents in chiral molecules by applying linearly polarized UV lasers. (a) A quantum optimal control simulation of a chiral molecular motor driven by femtosecond UV laser pulses was performed. A chiral molecule, (R)-2-methyl-cyclopenta-2, 4-dienecarboaldehyde, via the first singlet (nπ^*) excited state S_1 was treated as a molecular motor. Here, the aldehyde group acts as a roror part of the motor. Electric fields of femtosecond laser pulses driving both the regular rotation with a plus angular momentum and the reverse rotation with a minus one have been designed by using a global control method. The mechanisms of the unidirectional rotations have been clarified by analyzing the time-frequency-resolved spectrum of the electric fields of the optimal laser pulses and by resolving the temporal behaviors of rotational wave packets both in S_0 and S_1. The mechanism of the regular rotation is similar to that obtained by a conventional pump-dump pulse method. A new control mechanism has been proposed for the reverse rotation that cannot be driven by a simple pump-dump pulse method. (b) We have demonstrated that π electrons in a chiral aromatic molecule can be rotated along its ring by using a linearly polarized laser pulse. This indicates that molecular chirality can also be a source of directional ring currents because a nonhelical photon has no angular momentum. π-Electron dynamics reflects the asymmetry of the molecule itself. We have shown that the rotation direction of π electrons depends on the polarization direction of the applied pulse relative to the spatial configuration of each enantiomer on the basis of a three-level model analysis. We have performed optimal control simulations of π-electron rotation (ring current) in a six-membered control pulse at the maxima is a two-color laser field resonant with the quasi-degenerate states. The relative phase between the two frequency components is a crucial parameter for determination of the rotation direction of π-electrons at the end of control. Control scheme of π-electron rotation suggested by optimal control simulations is control in frequency domain by a two-color laser field while π-electron rotation can also be controlled in time domain by a central-peak laser pulse, e.g., a sin^2 or Gaussian pulse. In this research project, we have developed a fundamental basis of control of ultrafast electrons and vibrational states of chiral molecules by using linearly polarized lasers., 17350004
2005 - 2007