論文 - 山森 一人
-
Optimization of camera positions for taking all indoor sceneries by GA 査読あり
Y.Tominari, T.Nakagawa, K.Yamamori, I.Yoshihara, H.Takeda
Proc. Artificial Life and Robotics 2005 GS8-1 2005年2月
記述言語：英語 掲載種別：研究論文（国際会議プロシーディングス）
-
A Multi-modal Neural Network with Single-State Predictions for Protein Secondary Structure 査読あり
H.Zhu, I.Yoshihara, K.Yamamori, M.Yasunaga
The International Journal of Artificial Life and Robotics 8 ( 2 ) 168 - 173 2004年11月
記述言語：英語 掲載種別：研究論文（学術雑誌）
-
Gmdh-based Model Optimized by GA for Extracting Exon Regions from DNA Sequences
K.Ohta, I.Yoshihara, K.Yamamori, M.Yasunaga
Proc. Simulated Evolution and Learning 2004 SWP-8-2 2004年10月
記述言語：英語 掲載種別：研究論文（国際会議プロシーディングス）
-
GMDH-GA Hybrid Model Extracting Exon Region from DNA Sequences
Ohta Kouji, Yoshihara Ikuo, Yamamori Kunihito, Yasunaga Moritoshi
宮崎大學工學部紀要 33 289 - 294 2004年10月
記述言語：日本語 掲載種別：研究論文（学術雑誌） 出版者・発行元：宮崎大学
Abstract ###A model building method based on Group Method of Data Handling (GMDH) optimized by ###GA is developed for extracting exon regions. GMDH, that is originally a method to construct ###higher order polynomial model, is extended to constructing higher order logical model. ###The model built by proposed method is compared with Genetic Programming (GP)-based ###model as to the extraction rate of best, worst and average. The proposed method is superior to GP ###as to extraction rate of all for intron-exon boundary.
-
Onitani Toshiro, Yoshihara Ikuo, Yamamori Kunihito, Yasunaga Moritoshi
宮崎大學工學部紀要 33 301 - 306 2004年10月
記述言語：日本語 掲載種別：研究論文（学術雑誌） 出版者・発行元：宮崎大学
Abstract ###Finding transcription regulatory pattern of Dictyostelium discoideum (D. discoideum) is ###one of the most important tasks in genomics. Life cycle of D. discoideum is divided into ###vegetative stage and morphogenesis stage. The goal of this paper is to develop the technique for ###extracting the feature patterns in vegetative stage and morphogenesis stage. We calculate each ###score for candidate patterns ("aaaaaaaa"-"tttttttt") by matching these candidate patterns to ###DNA sequence of D. discoideum. The feature patterns of vegetative stage and morphogenesis ###stage are extracted by comparing scores of morphogenesis stage with those of vegetative stage. ###Experiments reveal that "cagcagca", "agcagcac", "gagagaga", "cagcatca", and ###"caccagca" appear in vegetative stage more than morphogenesis stage, and "cacaccc", ###"gtgtgtgt","acgactac","acacaccc",and "ctactact" appear in morphogenesis stage more than vegetative stage.
-
Extraction of Feature Patterns Embedded in Non-transcribed Region of Dictyostelium Discoideum
T.Onitani, I.Yoshihara, K.Yamamori, M.Yasunaga
Proc. Simulated Evolution and Learning 2004 SWP-8-1 2004年10月
記述言語：英語 掲載種別：研究論文（国際会議プロシーディングス）
-
High-speed Generation of Logic Function to Identify Exon-Intron Boundaries by Parallel GP
Yamamori Kunihito, Fujita Yuji, Yoshihara Ikuo, Aikawa Masaru
宮崎大學工學部紀要 33 343 - 348 2004年10月
記述言語：日本語 掲載種別：研究論文（学術雑誌） 出版者・発行元：宮崎大学
Abstract ###Genetic Programming (GP) is applied to various field. However, it is a problem that GP needs a ###lot of execution time for large scale problems such as genome informatics. In this paper, we focus on ###reducing computation time of GP. For this objective, we parallelize GP by island model on a personal ###computer (PC) cluster system to identify exon-intron boundaries in DNA sequences which is example ###of application. In addition, we described individuals as an one dimentional matrix to reduce the time of ###migration which is a process of island model. The paralellized GP achieved linear speedup ratio on each ###number of PC. Moreover, we show quality of solution to change on each number of PC.
-
Optimal Camera Layout to Take Pictures for Indoor Landscape Using GA
Yoshihara Ikuo, Nakagawa Takumi, Yamamori Kunihito, Takeda Haruo
宮崎大學工學部紀要 33 295 - 300 2004年10月
記述言語：日本語 掲載種別：研究論文（学術雑誌） 出版者・発行元：宮崎大学
ABSTRACT ###Taking pictures including all the walls and materials is necessary to make video-based CG of indoor ###landscapes. These pictures must have a variety of conditions such as a distance between camera and wall, ###overlap of pictures to compose CG and so on. This paper treats a building as two dimensional model and ###finds layout which camera can take all the parts of the walls in the building using Genetic Algorithms. ###We find such layouts and can minimize the number of cameras.
-
VHDLによる学習可能な階層型ニューラルネットワークのハードウェア実装
山森 一人, 石田 二郎, 吉原 郁夫
宮崎大學工學部紀要 33 355 - 359 2004年10月
記述言語：日本語 掲載種別：研究論文（学術雑誌） 出版者・発行元：宮崎大学
Neural network has been used many applications, such as pattern or image###recognition, robot control, optimization problem and so on. However, real-world###problems need large scale neural networks, and they lead enormous computation time###for training process of neural network. To reduce the· computation time, we try to###implement neural network in a FPGA device. In this paper, we discuss on the###performance of hardware neural network from the viewpoint of the processing speed###and the scale of the circuit. The trainable hardware neural network used 117,876 cells###in FPGA, and it could train the four training patterns in 800ns on the XOR problem.
-
Timetabling for Satisfying Professors' Requirements and Students' Desires using Genetic Algorithm
Yoshihara Ikuo, Sakaguchi Yoshiyuki, Yamamori Kunihito
宮崎大學工學部紀要 33 313 - 318 2004年10月
記述言語：日本語 掲載種別：研究論文（学術雑誌） 出版者・発行元：宮崎大学
Abstract ###Timetabling is to allocate the lectures in the time slot of a week, so as to fulfill various constraints. ###Timetable is re-created every year due to alternate professors, revision of a curriculum and so on. How-###ever, creating timetable is complicated because of a variety of constraints. This paper proposes a tech-###nique to create timetables using genetic algorithms. Timetabling problem is formulated as an optimiza-###tion problem which satisfies students' desires as much as possible and professors' requirements are ###considered as constraints. A university timetable is created using the proposed technique and compared ###with an actual timetable. The experiment shows that the proposed technique can create a timetable with ###same or higher quality as an actual timetable.
-
Optimization of Discrete Camera Position for Taking All Scenery Inside Buildings by GA
Yamamori Kunihito, Tominari Yusuke, Yoshihara Ikuo, Takeda Haruo
宮崎大學工學部紀要 33 361 - 365 2004年10月
記述言語：日本語 掲載種別：研究論文（学術雑誌） 出版者・発行元：宮崎大学
ABSTRACT ###Minimizing the number of cameras is search for optimal all location for taking all scenery inside ###buildings. It is necessary to take the conditions of a camera and walls into consideration. We proposed ###a GA-based method to minimize the number of cameras to take images all scenery inside buildings. ###We used an actual map, and experimented based on GA. It was able to ask for the number of a camera ###considered to be close to the minimum as a result.
-
山森 一人, 平山 政宏, 吉原 郁夫
宮崎大學工學部紀要 33 349 - 354 2004年10月
記述言語：日本語 掲載種別：研究論文（学術雑誌） 出版者・発行元：宮崎大学
Abstract ###In recent years, identification of persons becomes very important for security. ###Password has been generally used for this identification, however, persons ###sometimes forget their password, and sometimes it is stolen. Therefore a new ###technology is required for robust identification. In this paper, we use mosaic face ###images to recognize persons by self organizing map (SOM) proposed by Kohonen. ###In addition we evaluate the SOM based recognition by using of Euclid distance ###and city block distance. The SOM with the Euclid distance succeeded complete ###recognition, and the one with the city block distance achieved 96.7% recognition ###ratio.
-
Performance of Parallel Island Model with Dividing Population and Training-Set 査読あり
Y.Fujita, K.Yamamori, I.Yoshihara, M.Aikawa
Proc. Simulated Evolution and Learning 2004 STA-2-1 2004年10月
記述言語：英語 掲載種別：研究論文（国際会議プロシーディングス）
-
GA-Based Timetabling for Satisfying Both Professors and Students Requirements 査読あり
Y.Sakaguchi, I.Yoshihara, N.Koizumi, K.Yamamori, M.Yasunaga
Proc. Simulated Evolution and Learning 2004 STA-2-3 2004年10月
記述言語：英語 掲載種別：研究論文（国際会議プロシーディングス）
-
Hybrid GA using Edge Assembly Crossover and Lin-Kernighan Heuristic 査読あり
M.Sato, I.Yoshihara, D.H.Nguyen K.Yamamori, M.Yasunaga
Proc. Simulated Evolution and Learning 2004 STA-8-1 2004年10月
記述言語：英語 掲載種別：研究論文（国際会議プロシーディングス）
-
English Pronunciation Reasoning Considering Frequency of Appearance of Phonemes by Neural Network 査読あり
Y.Higashi, I.Yoshihara, H.Zhu, K.Yamamori, M.Yasunaga
Proc. Simulated Evolution and Learning 2004 SWA-8-1 2004年10月
記述言語：英語 掲載種別：研究論文（国際会議プロシーディングス）
-
Optimal Camera Layout to Take Pictures for Indoor Landscape Using GA 査読あり
T.Nakagawa, Y.Tominari, I.Yoshihara, K.Yamamori, H.Takeda
Proc. Simulated Evolution and Learning 2004 STA-2-2 2004年10月
記述言語：英語 掲載種別：研究論文（国際会議プロシーディングス）
-
Multi-Modal Neural Networks for Symbolic Sequence Pattern Classification
ZHU Hanxi, YOSHIHARA Ikuo, YAMAMORI Kunihito, YASUNAGA Moritoshi
IEICE transactions on information and systems 87 ( 7 ) 1943 - 1952 2004年7月
記述言語：日本語 掲載種別：研究論文（学術雑誌） 出版者・発行元：一般社団法人電子情報通信学会
We have developed Multi-modal Neural Networks (MNN) to improve the accuracy of symbolic sequence pattern classification. The basic structure of the MNN is composed of several sub-classifiers using neural networks and a decision unit. Two types of the MNN are proposed: a primary MNN and a twofold MNN. In the primary MNN, the sub-classifier is composed of a conventional three-layer neural network. The decision unit uses the majority decision to produce the final decisions from the outputs of the sub-classifiers. In the twofold MNN, the sub-classifier is composed of the primary MNN for partial classification. The decision unit uses a three-layer neural network to produce the final decisions. In the latter type of the MNN, since the structure of the primary MNN is folded into the sub-classifier, the basic structure of the MNN is used twice, which is the reason why we call the method twofold MNN. The MNN is validated with two benchmark tests: EPR (English Pronunciation Reasoning) and prediction of protein secondary structure. The reasoning accuracy of EPR is improved from 85.4% by using a three-layer neural network to 87.7% by using the primary MNN. In the prediction of protein secondary structure, the average accuracy is improved from 69.1% of a three-layer neural network to 74.6% by the primary MNN and 75.6% by the twofold MNN. The prediction test is based on a database of 126 non-homologous protein sequences.
-
Multi-modal neural networks for symbolic sequence pattern classification
Zhu H., Yoshihara I., Yamamori K., Yasunaga M.
IEICE Transactions on Information and Systems E87-D ( 7 ) 1943 - 1952 2004年7月
記述言語：日本語 掲載種別：研究論文（学術雑誌） 出版者・発行元：IEICE Transactions on Information and Systems
We have developed Multi-modal Neural Networks (MNN) to improve the accuracy of symbolic sequence pattern classification. The basic structure of the MNN is composed of several sub-classifiers using neural networks and a decision unit. Two types of the MNN are proposed: a primary MNN and a twofold MNN. In the primary MNN, the sub-classifier is composed of a conventional three-layer neural network. The decision unit uses the majority decision to produce the final decisions from the outputs of the sub-classifiers. In the twofold MNN, the sub-classifier is composed of the primary MNN for partial classification. The decision unit uses a three-layer neural network to produce the final decisions. In the latter type of the MNN, since the structure of the primary MNN is folded into the sub-classifier, the basic structure of the MNN is used twice, which is the reason why we call the method twofold MNN. The MNN is validated with two benchmark tests: EPR (English Pronunciation Reasoning) and prediction of protein secondary structure. The reasoning accuracy of EPR is improved from 85.4% by using a three-layer neural network to 87.7% by using the primary MNN. In the prediction of protein secondary structure, the average accuracy is improved from 69.1% of a three-layer neural network to 74.6% by the primary MNN and 75.6% by the twofold MNN. The prediction test is based on a database of 126 non-homologous protein sequences.
-
Multi-modal Neural Networks for Symbolic Sequence Classification 査読あり
H.Zhu, I.Yoshihara, K.Yamamori, M.Yasunaga
IEICE Transactions on Information Systems E87-D ( 7 ) 1943 - 1952 2004年7月
記述言語：英語 掲載種別：研究論文（学術雑誌）