Publisher：ACM  

researchmap

Other Link： http://dblp.uni-trier.de/db/conf/iccae/iccae2018.html#conf/iccae/AbeMH18

Language：Japanese   Publisher：Information Science and Technology Association, Japan  

<p>As a pre-award activity of URA, sending e-mail of research grant to researchers within the university is widely practiced. There are many people who want to know only the public offerings related to the research field of researchers themselves. Therefore, we developed a recommendation system -Galette- that extracts researchers who match the theme of the public offering. As researcher data, we used research projects acquired from the KAKEN database to create feature vectors by natural language processing. We implemented a system that creates the feature vector representing the public offering, ranks them by cosine similarity. We evaluated the recommendation results through cooperation of URA and got good results. We aim to develop a more useful system for URA.</p>

DOI： 10.18919/jkg.68.6_298

CiNii Books

researchmap

Language：English  

researchmap

Language：English  

researchmap

Publisher：ACM  

researchmap

Other Link： http://dblp.uni-trier.de/db/conf/icsps/icsps2016.html#conf/icsps/MoritaAH16

DOI： 10.18494/SAM.2016.1181

researchmap

Language：English   Publishing type：Research paper (international conference proceedings)   Publisher：IEEE  

The impression of a video changes depending on the audio and visual content of the video. Adding appropriate background music to a video is an important process for making the video more impressive. In this paper, we present a system called OtoPittan which recommends background music for video based on the valence and arousal model. As input for the system, first a user registers a video with no background music, and then inputs a desired impression by setting the valence and the arousal level. The system recommends music clips as candidates of background music for the video. Recommended music clips are determined by minimizing the difference between the desired impression and the predicted impression calculated from the audio features of each candidate music and the visual features of the registered video. We have confirmed that with the proposed system users can quickly create videos giving desired impressions.

DOI： 10.1109/ISM.2016.101

Web of Science

researchmap

Language：English   Publishing type：Research paper (scientific journal)   Publisher：MYU, SCIENTIFIC PUBLISHING DIVISION  

Heart rate estimation during exercise with photoplethysmographic (PPG) sensors is a difficult problem because motion artifacts (MAs) contaminate PPG signals. In this paper, we propose a method of accurately estimating heart rate during exercise from a PPG signal by considering exercise intensity. The exercise intensity is estimated from acceleration signals simultaneously recorded with the PPG signal. The proposed method dynamically limits the estimation range of heart rate depending on the exercise intensity, resulting in an accurate heart rate estimation during exercise. From the experimental results, we have confiimed that the proposed method can estimate heart rate more accurately than conventional methods.

DOI： 10.18494/SAM.2016.1184

Web of Science

researchmap

Language：English   Publishing type：Research paper (international conference proceedings)   Publisher：IEEE  

This paper presents a system for similarity retrieval of 3D models. In the system, a 3D sketch created with clay is used as a query. The clay sketch is digitized into a point cloud with a desktop 3D scanner. To compensate for the lack of scanning accuracy, the digitized model is refined by interpolation and decimation of data points. The effectiveness of the proposed system is experimentally evaluated by comparing the query-by-clay sketch method with other methods. The results have shown that the retrieval performance of the query-by-clay sketch method is superior to conventional methods.

DOI： 10.1109/ISM.2016.125

Web of Science

researchmap

Language：English   Publishing type：Research paper (scientific journal)   Publisher：Atlantis Press  

Object extraction has a crucial role in various visual semantic scenarios. In this paper, we propose a system for object extraction which can deal with an object partly occluded by other objects. In order to estimate the hidden part of the occluded object, the system combines different types of contour completion methods such as curve completion and symmetry completion. The system is composed of two modules: contour discrimination and contour completion. The contour discrimination module separates the contour of the occluded object from the contours of the occluding objects. To the contour of the occluded object, the contour completion module applies different types of contour completion algorithms to generate various completion patterns. Based on the perceptual model of figural goodness, the system computes the figural goodness scores of the generated patterns, where the simplicity and symmetricity are evaluated. Finally, the system outputs the pattern having the highest score as a result of the object extraction. From the experimental results, we have confirmed that the figural goodness model works effectively for extracting partly occluded objects.

DOI： 10.2991/ijndc.2015.3.3.6

Scopus

researchmap

Language：English   Publishing type：Research paper (scientific journal)   Publisher：Atlantis Press  

While many interfaces for gaze writing (text entry by eye gaze) are available today, these interfaces are just a replacement of a keyboard. For more efficient gaze writing, this paper proposes a novel gaze writing system which has two kinds of gaze writing interfaces: (1) an interface for text entry and (2) an interface for text copy and paste. The interface for text entry was selected from conventional interfaces based on experiments of workload evaluation of beginner users. The interface for text copy and paste, which is newly developed in our study, enables a user to specify a text range and paste the text to a desired position with only gaze interaction. To overcome the lack of accuracy of gaze pointing, the proposed system adopts the model of two step cursor control, which processes initial cursor setting and adjustment in turn. The experimental results have shown that the proposed interface makes gaze writing more efficiently.

DOI： 10.2991/ijndc.2015.3.3.7

Scopus

researchmap

Language：Japanese   Publishing type：Research paper (scientific journal)   Publisher：Japanese Society for Artificial Intelligence  

This paper proposes a movie recommendation system using movie reviews on the Web. The system receives a movie review from a user, estimates the user’s interests on movies from the review, and provides other persons’ reviews to the user based on interest matching. This paper assumes that user’s interests on movies appear on words which are positively or negatively evaluated in a review. Under the assumption, the system detects such words from the user’s review, and chooses other persons’ reviews to recommend in which the detected words are positively evaluated. Experimental results have shown that more than 1/3 of recommended reviews can motivate users to watch the movies mentioned in the reviews. In addition, the results have indicated that combining the proposed recommendationmethod and a conventional TF-IDF based recommendation method is important for more efficient recommendation.

DOI： 10.1527/tjsai.30.102

Scopus

researchmap

Language：English   Publishing type：Research paper (international conference proceedings)   Publisher：IEEE  

Gaze writing (text entry by eye gaze) is an indispensable way of communication for motor-disabled people. While many interfaces for gaze writing are available today, the purpose of these interfaces is just a replacement of a keyboard. In order for users to communicate more efficiently, this paper introduces an interface for text copy and paste into a gaze writing system. The interface enables a user to specify a copy range and paste text to a desired position with only gaze interaction. To overcome the lack of accuracy of gaze pointing, the proposed system adopts a model of two step cursor control, which processes initial cursor setting and adjustment in turn. The experimental results have shown that the proposed interface makes gaze writing more efficiently.

DOI： 10.1109/ICIS.2015.7166581

Web of Science

researchmap

Language：English   Publishing type：Research paper (international conference proceedings)   Publisher：IEEE  

In this paper, we propose an object decomposition system which can deal with partly occluded objects. In order to develop shape-independent object decomposition, the system combines different types of contour completion methods, namely, line completion, curve completion, line-symmetry completion and rotational symmetry completion. The system evaluates figural goodness of each completion pattern generated by these completion methods from the aspect of simplicity and symmetricity, and then selects the best one to output. From the experimental results, we have confirmed that the proposed system can correctly decompose objects from an input image.

DOI： 10.1109/ICIS.2015.7166611

Web of Science

researchmap

Language：English   Publishing type：Research paper (international conference proceedings)   Publisher：IEEE  

In this paper, we propose a framework for extending existing matting methods to actualize more reliable alpha estimation. The key idea of the framework is integration of multiple alpha maps based on their reliabilities. In the proposed framework, the given input image is converted into multiple grayscale images having various luminance appearances. Then, alpha maps are generated corresponding to these grayscale images by utilizing an existing matting method. At the same time reliability maps (single channel images visualizing the reliabilities of the estimated alpha values) are generated. Finally, by combining alpha maps having the highest reliabilities in each local region, one reliable alpha map is generated. The experimental results have shown that reliable alpha estimation can be actualized by the proposed framework.

DOI： 10.1109/ISM.2015.28

Web of Science

researchmap

Language：English   Publishing type：Research paper (international conference proceedings)   Publisher：IEEE  

This paper reports improvement of indirect matching, which is a fast CBMIR (content-based music information retrieval) framework proposed in our previous study. Indirect matching achieves fast retrieval by combining offline search with representative queries and online quick similarity estimation based on the results of the offline search. We have found that the retrieval accuracy of indirect matching decreases when representative queries have little variation. This paper proposes a method for selecting representative queries having wide variation. To ensure wide variation between representative queries, the proposed method combines MDS (multi-dimensional scaling) and Ward's clustering. Experimental results have shown that the retrieval accuracy of indirect matching can be stabilized by the proposed method.

DOI： 10.1109/ISM.2015.29

Web of Science

researchmap

DOI： 10.1142/S1793351X14400030

researchmap

Language：English   Publishing type：Research paper (international conference proceedings)   Publisher：IEEE  

This paper presents a fast content-based music information retrieval method. The high computational cost of similarity evaluation based on musical features between a pair of music clips is a crucial problem especially for searching large music database. To reduce the computational time in similarity evaluation process, the proposed method adopts an approach called indirect matching. In the approach, a small number of music clips called representative queries, which are randomly selected from a database, are used for fast computation. As an offline process, the similarities of each music clip in the database to the representative queries are recorded as a similarity table. In the online phase, the similarity between the actual query (the music clip given by a user) and each music clip in the database is quickly estimated by referring the similarity table. Experimental results have shown that the execution time of retrieval can be greatly reduced by the indirect matching without much deterioration of retrieval accuracy.

Web of Science

researchmap

Other Link： http://dblp.uni-trier.de/db/conf/eusipco/eusipco2014.html#conf/eusipco/HayashiIY14

Language：English   Publishing type：Research paper (international conference proceedings)   Publisher：IEEE  

This paper proposes a method for contour completion of a skew-symmetry object which is partly occluded by other objects. The proposed method assumes that skew symmetry of a shape can be measured by correlation with its reflection on a nonorthogonal plane. Under the assumption, the method estimates a skew-symmetry axis and a skew-transverse angle, and then reflects the contour to the estimated axis considering the skew-transverse angle. Using the reflected contour as a patch, the occluded part of the contour is restored. However, due to the information loss caused by occlusion, the reflected contour slightly moves from the correct position. The method moves the reflected contour within a small range to smoothly connect it with the original contour. We experimentally confirmed the effectiveness and limitation of the proposed method by comparing the proposed method with the human perception.

DOI： 10.1109/ISM.2014.15

Web of Science

researchmap

Language：English   Publishing type：Research paper (international conference proceedings)   Publisher：IEEE  

This paper presents workload evaluation of gaze-writing systems. As a method for evaluating user's workload of a gaze-writing system, we used NASA-TLX, a questionnaire-based method for workload evaluation. We partly modified the NASA-TLX questionnaire to suit them for users of gaze-writing systems. We experimentally compared the workloads of users of two traditional gaze-writing systems QWERTY and Dasher. The experimental results showed that the workload of QWERTY tended to be lower than Dasher.

DOI： 10.1109/ISM.2014.18

Web of Science

researchmap

Language：Japanese   Publisher：一般社団法人電子情報通信学会  

CiNii Books

researchmap

Publisher：IEEE Computer Society  

DOI： 10.1109/ICIS.2013.6607897

researchmap

Language：English   Publishing type：Research paper (international conference proceedings)   Publisher：IEEE Computer Society  

This paper proposes a collaborative filtering based spot recommendation system seamlessly available in user's home areas and away areas. The system recommends spots to a user which are highly evaluated by other users whose preference in spot genre selection is similar to the user's one and whose home areas cover the user's current location. As an application of the proposed system, we have implemented a restaurant recommendation system. Experimental results have shown that the superiority of the proposed system to an existing spot recommendation system. © 2013 IEEE.

DOI： 10.1109/ICIS.2013.6607896

Scopus

researchmap

Language：English   Publishing type：Research paper (international conference proceedings)   Publisher：IEEE  

This paper proposes a system for completing the contour of an object which is partly occluded by other objects. The proposed system classifies objects into rotational symmetry objects, line symmetry objects and the other objects. To a rotational symmetry object, the system estimates the angle and center of rotation of the object focusing on the periodicity of the curvature change of the contour. Rotating the object with the estimated angle and center of rotation, the system reconstructs the occluded part of the contour. To a line symmetry object, the system detects line symmetry axes. The occluded part is completed by folding the shape to the line symmetry axes. To other shapes, the system uses a spline interpolation to complete the contour. We have experimentally confirmed the effectiveness and limitations of the proposed method by comparing the proposed method with human perception.

DOI： 10.1109/ICIP.2013.6738728

Web of Science

researchmap

Language：English   Publishing type：Research paper (international conference proceedings)   Publisher：IEEE  

Adding BGM (background music) to a video is an important process in video creation because BGM determines the impression of the video. We model impression estimation of a video as mappping from computer-mesurable audio and visual features to impression degrees. As an application of impression estimation of a video, we propose OtoPittan, a system for recommending BGM for helping users to make impressive videos. OtoPittan regards the problem of selecting BGM from a music collection as a partial inverse problem of the impression estimation. That is, to an inputted video and desired impression, BGM which produces a good match to the desired impression when adding it to the inputted video is recommended. As implementation ways of impression estimation of a video, we use a static user model and a dynamic user model. The first model statically constructs a mapping function learnt from training data. The second model dynamically optimizes a mapping function through user interaction. Experimental results have shown that the static user model has high estimation accuracy and the dynamic user model can efficiently performs optimization without much user interaction.

DOI： 10.1109/ISM.2013.25

Web of Science

researchmap

Language：English   Publishing type：Research paper (international conference proceedings)   Publisher：IEEE  

This paper presents a fast similarity retrieval method for vector images. To reduce the computational cost of similarity matching, the proposed method uses pre-calculation results of similarity matching, which are obtained in advance by matching DB images with previously selected images called representative queries. At runtime the proposed method just matches the actual query (the user-inputted query) and the representative queries. Comparing the similarities with the precalculated similarities, the proposed method quickly estimates the actual similarities of DB images to the actual query. Experimental results have shown that the retrieval time is greatly reduced by the proposed method without much deterioration of retrieval accuracy.

DOI： 10.1109/ISM.2013.95

Web of Science

researchmap

Language：English   Publishing type：Research paper (international conference proceedings)   Publisher：ACM  

According to the Gestalt principles, this paper presents a method for recognizing grouping areas of good continuity in trademark images. The purpose of this investigation is to enhance performances of content-based image retrieval by mirroring human perception. On recognizing grouping areas on good continuity, the physical factors suggested by the Gestalt Psychology are: proximity between couples of components in a set of components in one image and shape similarity between these components. Considering the factors, the proposed method measures proximity between every couple of components and judges whether the couple could be a part of the grouping area using a support vector machine. Then, the proposed method extracts features on shape similarity from every component and measures difference of the features between every couple of components. After then, using the difference, the couple is judged as well as the case of proximity. The learning data in the SVM are determined by results of questionnaire on good continuity for 75 sample images. To examine performance of the proposed method, another 75 trademark images have been tried and experimental results have shown correspondence ratios between grouping patterns output by the proposed method and results of the questionnaire have been more than 85.41%. © 2012 ACM.

DOI： 10.1145/2425836.2425895

Scopus

researchmap

Other Link： http://dblp.uni-trier.de/db/conf/ivcnz/ivcnz2012.html#conf/ivcnz/BefaneAH12

Language：Japanese  

CiNii Books

J-GLOBAL

researchmap

Language：English   Publishing type：Research paper (international conference proceedings)   Publisher：IEEE Computer Society  

This paper presents a fast and accurate object search algorithm based on template matching. Template matching is an important technique within the field of image processing for searching object region from images which matches a template. Template matching has a problem that existence of background in a template causes search error. To avoid such error, in the proposed object search, a procedure called silhouette masking is newly added. In addition, to reduce computational cost, the framework of active search is used. Experimental results have shown that the proposed search has higher accuracy than the simple active search and the brute force search because the silhouette masking works efficiently. Besides, comparison between the proposed search and the brute force search has shown that computational cost can be greatly reduced by the proposed search. © 2010 IEEE.

DOI： 10.1109/ICIS.2010.115

Scopus

researchmap

Language：Japanese   Publisher：情報処理学会  

CiNii Books

J-GLOBAL

researchmap

Publisher：IEEE Computer Society  

DOI： 10.1109/ICDMW.2009.58

researchmap

Language：Japanese  

CiNii Books

researchmap

Language：Japanese   Publisher：日本ソフトウェア科学会  

DOI： 10.11309/jssst.26.4_138

J-GLOBAL

researchmap

Language：Japanese   Publisher：一般社団法人 電気学会  

DOI： 10.1541/ieejsmas.129.173

J-GLOBAL

researchmap

Language：English   Publishing type：Research paper (scientific journal)  

This paper presents a method for discriminating between personal and non-personal web pages. The method can support surveys of personal opinions about products and services. In the proposed method, subjective expressions are extracted from pages and then the pages are scored by quantitatively evaluating the subjectivity in the pages. We have evaluated performances of the proposed method using 1200 web pages collected from four categories of product, tourist spot, restaurant, and movie. Comparing the performances of the proposed method with categorisations by a general search engine, we have confirmed that the performances have been significantly better in every category. Copyright © 2009, Inderscience Publishers.

DOI： 10.1504/IJBIDM.2009.025411

Scopus

researchmap

Publisher：IEEE Computer Society  

DOI： 10.1109/WAINA.2008.52

researchmap

Language：English   Publishing type：Research paper (international conference proceedings)   Publisher：IEEE  

A novel method for content-based 2D vector image retrieval is proposed in this paper. To retrieve vector images, first, the proposed method extracts WFPs (Weighted Feature Points) from vector images. In this extraction process, the method directly accesses the parameters of Bezier curves defining the shapes of vector images and computes WFPs of the images according to these parameters. Second, to a pair of vector images, the method evaluates similarity by matching WFPs using PTD (Proportional Transfer Distance). In the proposed method, vector images are not rasterized, which enables real-time retrieval. As a prototyping, we have implemented a system for sketch-based vector image retrieval. In the system, a sketch is used as a query. To bridge the representation gap between sketches and vector images in a database, the strokes in sketches are approximated with Bezier curves as a preprocess. Using the system and a database containing 1,120 logomarks, we have confirmed the retrieval performance and have discussed the limits of the proposed method.

DOI： 10.1109/ICIP.2008.4711916

Web of Science

researchmap

Publisher：IASTED/ACTA Press  

researchmap

Language：Japanese   Publisher：一般社団法人情報処理学会  

This paper reports on a prototype system RE:Why for extracting answers of "Why" type questions from WEB. We have proposed an algorithm for finding an answer in a page and a scoring method for extracted answers, and have implemented a prototype system using these ideas We have evaluated the effectiveness of our algorithm using 10 questions and have confirmed that the algorithm has 31.3% precision and 50.4% recall. We have conducted experiments using 20 questions and have shown that our system output the correct answer to 16 questions of 20 questions.

CiNii Books

researchmap

Language：Japanese   Publisher：一般社団法人情報処理学会  

This paper presents a novel object extraction from 2D vector images. Although various object extractions from raster images have been proposed, none of them has a mechanism for handling vector images. Therefore, in order to extract object regions from vector images using previous methods, it is necessary to convert vector images into raster images. However, various mertis of vector image are discared by rasterization. In this study, we have proposed a novel object extraction method using merits of vector images. The proposed method analyzes relations between individual primitives of vector images and the regions defined by these primitives. Accoring to the analysis, a set of primitives defining object region is detected. We have compared accuracy and speed performance between the proposed method and ACM (Active Contour Model), which is a traditional raster-based object extraction method. Experimental results have shown that the proposed method has higher accuracy and speed performance than ACM. In addition, we have confirmed the proposed method has room for improvement in speed performance.

CiNii Books

researchmap

DOI： 10.20965/jaciii.2007.p0061

researchmap

Language：English   Publishing type：Research paper (scientific journal)   Publisher：一般社団法人 画像電子学会  

Contents of sports tickers in broadcasted sports videos could be efficient as queries in analyzing the video contents. However, since sports tickers are displayed with various colors and brightness, it is difficult to recognize inner objects of the sports tickers only according to the brightness distribution. Besides, in order to recognize what the objects represent, it is indispensable to investigate the change of these objects according to time traveling. In this paper, the proposed method can recognize locations of the out ticker, score tickers, and runner markers in the sports ticker of broadcasted baseball videos analyzing the brightness histograms obtained from every of grayscale and color images, and the change of the sports ticker with the correlation coefficient. Experimental results with four broadcasted baseball videos have shown that the proposed method can exactly work in the case when the out ticker is represented by a letter. © 2007, The Institute of Image Electronics Engineers of Japan. All rights reserved.

DOI： 10.11371/iieej.36.701

Scopus

researchmap

Language：English   Publishing type：Research paper (international conference proceedings)   Publisher：IEEE COMPUTER SOC  

This paper proposes a novel method for vector image segmentation as the first step in developing a content-based vector image retrieval system. Structure of a vector image can be represented as a tree in which each node is assigned each object region in the vector image and each link represents the inclusion relation between two object regions. In order to generate such trees, the method separates object regions from background by detecting figures defining boundary between object regions and background. The proposed method finds object regions before rasterizing, which is an essential difference from existing object separation methods. We have evaluated the effectiveness of the proposed object separation on 40 test vector images by comparing manual object separation. The experimental results have shown that the proposed method has a high performance which is comparable to manual object separation.

DOI： 10.1109/CIT.2007.107

Web of Science

researchmap

Language：Japanese   Publisher：情報処理学会  

Genetic Algorithms (GA) are optimization methods inspired by evolutionary mechanisms. GA fall into states which can't improve solutions (evolutionary stagnation) because indivisuals are always evoluated in order to improve. GA have a difficulty problem to obtain the optimal solutions. As a way in order to dissolve this problem, the diversity maintenance of population, however, the problem occurs, which convergence speed becomes slow. This paper proposes a novel GA to avoid evolutionary stagnation by corrupting the solutions and also convergence speed does't become slow, by using improving population which improves the solutions, and corrupting population which corrupts the solutions. This paper also shows that good results compared with traditional GA, Minimal Generation Gap (MGG) as generation alternation and Dual Individual Distributed Genetic Algorithm (Dual DGA) on vehicle routing problems with time windows.

CiNii Books

J-GLOBAL

researchmap

Publisher：IASTED/ACTA Press  

researchmap

Language：Japanese   Publisher：一般社団法人電子情報通信学会  

CiNii Books

J-GLOBAL

researchmap

Language：Japanese   Publisher：日本知能情報ファジィ学会  

This paper proposes and evaluates a method for extracting personal web pages from a large number of unclassified web pages. We can use the method as a content filtering method for reputation searches. To extract personal pages from unclassified pages, the method focuses on four kinds of text features that appear at a personal page. The method quantitatively measures these features for each page and divides the pages into plural groups using k-means clustering based on the results of the measuring. From the groups the method finds groups that consist of personal web pages. We have evaluated the search performance of the method by measuring precisions. Experimental results have shown the average performance of the method is 2.1-times higher than the one of a keyword-based search engine.

DOI： 10.3156/jsoft.18.161

CiNii Books

J-GLOBAL

researchmap

Language：Japanese   Publisher：一般社団法人情報処理学会  

This paper reports on MineBlog: a recommender system of blogs. Posting a blog-article to the system, the system produces attractive blog-articles to the user and supports the user to discover his/her new interests. The system extracts some blog-articles for recommendation from a database by scoring with three kinds of criteria - relevance, difference and topicality. Relevance is a criterion for measuring similarity between an article registered in a database and the user's posted article. Difference is a criterion for measuring dissimilarity between the articles. Topicality is a criterion for measuring whether an article mentions current topics. We experimentally examine the performance of MineBlog. Experimental results show that one of two recommended articles arouse user's new interests.

CiNii Books

researchmap

Language：English   Publishing type：Research paper (scientific journal)   Publisher：SPRINGER-VERLAG BERLIN  

We present a system for retrieving web pages which contain product reputations. We assume that product reputations are contained in personal pages and message board pages, and propose the system to retrieve those web pages. We illustrate that the system can distinguish those web pages automatically from other pages using six kinds of features extracted from a web page. We have evaluated the performance of the proposed system by computing precision and recall measures.

DOI： 10.1007/978-3-540-30134-9_26

Web of Science

researchmap

Language：Japanese   Publisher：情報処理学会  

Genetic algorithms (GAs) can be applied to problems in various fields because there are not restricted in terms of solution expression or evaluation, however, GAs have a serious disadvantage in that they cannot be applied to changing environments. In this paper, a novel type GA for optimization under a dynamic environment is proposed. The method deal with (1) change of fitness landscape shape, (2) change of effect on phenotype at each locus and (3) change of locus-groups which organize linkages. Search scale control and block identification mechanisms, which are for treating these changes, is explained. Effectiveness of the method is confirmed by three simulations using changing knapzack problems and changing trap function optimization problems.

CiNii Books

researchmap

Language：Japanese   Publisher：電子情報通信学会  

CiNii Books

researchmap

Language：Japanese   Publisher：電子情報通信学会  

CiNii Books

researchmap

Language：Japanese  

DOI： 10.1527/tjsai.16.202

CiNii Books

researchmap

Language：Japanese   Publishing type：Research paper (scientific journal)   Publisher：一般社団法人 人工知能学会  

A hypothetical reasoning, which can find an explanation for a given set of observation by assuming some hypothetical sets, is a useful knowledge processing framework because of its theoretical basis and usefulness for practical problems such as diagnostics, scheduling, and design. It is, however, known to be computationally very expensive for large problems because it must deal with incomplete knowledge. Predicate-logic allows powerful and compact knowledge representation compared to propositional-logic. Efficient methods to proceed reasoning have been developed for propositional-logic. However it would be desirable to develop similar methods for predicate-logic, but this has so far proved difficult. "KICK-HOPE" is one of hypothetical reasoning systems for predicate-logic. KICK-HOPE, which is based on a deductive database technique called the QSQR method and a basic fast hypothetical reasoning system called ATMS(Assumption-based Truth Maintenance System), made advances in the speed of reasoning. However, for a practical knowledge base, 'the inconsistent process' and 'the subsumption process' still consume much calculation time. In this paper, we investigate an e.cient method of the inconsistent and subsumption processes of KICK-HOPE. First, we show most of these processes are vain and omissible. Second, to omit vain processes, we introduce bit-vector(check-vector) to represent what hypotheses include in hypothetical sets. And, we improve that vain processes will be reduced to check the check-vector. Finally, we examine the proposed system with two examples, a diagnostic problem of a logic circuit and a scheduling problem. It is shown to be more efficient than the original KICK-HOPE for all problems.

DOI： 10.1527/tjsai.16.202

Scopus

researchmap

Language：Japanese   Publisher：電気学会 = Institute of Electrical Engineers of Japan  

DOI： 10.1541/ieejsmas.120.533

researchmap

Language：Japanese   Publisher：一般社団法人電子情報通信学会  

CiNii Books

researchmap

Language：Japanese   Publisher：人工知能学会  

CiNii Books

researchmap

Language：Japanese   Publisher：人工知能学会  

CiNii Books

researchmap

researchmap

researchmap

Language：Japanese  

J-GLOBAL

researchmap

researchmap

researchmap

researchmap

Language：Japanese  

J-GLOBAL

researchmap

Language：Japanese  

J-GLOBAL

researchmap

researchmap

researchmap

researchmap

researchmap

researchmap

researchmap

researchmap

Language：Japanese  

J-GLOBAL

researchmap

researchmap

researchmap

researchmap

researchmap

researchmap

researchmap

researchmap

researchmap

researchmap

researchmap

researchmap

researchmap

researchmap

researchmap

researchmap

researchmap

researchmap

researchmap

Language：Japanese  

J-GLOBAL

researchmap

Language：Japanese  

J-GLOBAL

researchmap

Language：Japanese   Publisher：The Institute of Electronics, Information and Communication Engineers  

In this paper, we propose a method for completing the contour of an object which is partially occluded by another object. We assume that occluded objects have a line symmetric axis. In order to detect the axis, the proposed method folds back the object along a line and counts the overlapped pixels in the contour region. The line along which the contour is mostly overlapped is detected as the line symmetric axis. By folding back the object to the detected axis, the absent part in the contour is completed. As an experiment, we have compared the results of completion by the proposed method and humans. From the experimental results, we have discussed the effectiveness and limitation of the proposed method.

CiNii Books

J-GLOBAL

researchmap

Language：Japanese   Publisher：The Institute of Electronics, Information and Communication Engineers  

The grammar of the film language formalizes the patterns of camera works and film editing techniques used in movies. This research proposes a method for detecting two persons' conversation scenes from a movie according to the patterns of camera works and screen composition in the grammar of the film language. In such conversation scenes, one person is in the left or right half region in a shot, and another person is in the opposite side region in another shot. According to these characteristics, the proposed method detects conversation scenes. In the method, faces and eyes are detected first. Based on the information about the faces and eyes, the system detects conversation scenes. As an experiment, we have tried detection of conversation scenes from five movies, and have confirmed the effectiveness and limitation of the proposed method.

CiNii Books

J-GLOBAL

researchmap

Language：Japanese  

J-GLOBAL

researchmap

Language：Japanese  

J-GLOBAL

researchmap

Language：Japanese  

J-GLOBAL

researchmap

Language：Japanese  

J-GLOBAL

researchmap

Language：Japanese  

J-GLOBAL

researchmap

Language：Japanese  

J-GLOBAL

researchmap

Language：Japanese  

J-GLOBAL

researchmap

Language：Japanese  

J-GLOBAL

researchmap

Language：Japanese  

J-GLOBAL

researchmap

Language：Japanese   Publisher：Information Processing Society of Japan (IPSJ)  

This paper reports on ARIBROS: a system for recommending bloggers to a user. When the user posts a blog to the system, the system estimates the user interests and recommends bloggers who have similar interests to the user. The user can discover his/her new interests by the recommendation. The system has a feature that transition of bloggers' interests is represented with the metaphor of motion of ants. Bloggers are represented as ants. Keywords which express bloggers' interests are represented as foods. Ants move between foods according to transition of bloggers' interests. The user can discover new interests by observing motion of ants.

CiNii Books

J-GLOBAL

researchmap

Language：Japanese  

J-GLOBAL

researchmap

Language：Japanese  

J-GLOBAL

researchmap

Language：Japanese  

J-GLOBAL

researchmap

Language：Japanese  

J-GLOBAL

researchmap

Language：Japanese   Publisher：The Institute of Electronics, Information and Communication Engineers  

Previous researches on object tracking have improved tracking accuracy. However, it is difficult to track an object region in a video with no error. When a tracking error occurs, a user must re-extract the object region by hand. This paper proposes an object tracking method which user can easily re-extract the object region when a tracking error occurs. The proposed method applies graph-cut-based object extractions to object tracking in a video. Graph-cut-based object extraction separates an object region from the background using two kinds of lines which are drawn in the object region and the background region by a user. Using these two kinds of lines, the proposed method extracts an object region at the first frame in a video. In addition, the method extracts an object region in each frame after the first frame by tracking the lines the user has given. Even though a tracking error is occurred at some frame, user can easily repair the error by adding new lines and erasing existing lines at the frame. We have compared tracking accuracy between the proposed method and a traditional tracking method and have discussed the efficiency of the proposal.

CiNii Books

J-GLOBAL

researchmap

Language：Japanese   Publisher：Information Processing Society of Japan (IPSJ)  

This paper presents a prototype system for content-based vector image retrieval. The targets of the system are abstract figures such as logotypes. The system represents a vector image as a set of weighted feature points and defines similarity between two vector images using PTD (Proportional Transportation Distance) between the weighted feature points of the two images. The system extracts points which have large curvature change as feature points and varies weights given to the feature points depending on curvature change. We have evaluated the efficiency of the system using about 1100 vector images. The experimental results have shown that the system has high precision on search for vector images whose outlines are composed of points which have high curvature changes.

CiNii Books

J-GLOBAL

researchmap

Language：Japanese   Publisher：日本ソフトウェア科学会  

There have been many studies on question-answering systems on the Web. Some of them try to find the answers to &amp;ldquo;Who&amp;rdquo;, &amp;ldquo;When&amp;rdquo; and &amp;ldquo;Where&amp;rdquo; type questions. However, there are few studies on practical systems which allow &amp;ldquo;Why&amp;rdquo; type questions. Because &amp;ldquo;Why&amp;rdquo; type questions need sentences for the answers and traditional and typical methods give the answers as a phrase or a word. Therfore we have proposed and implemented a prototype system &amp;ldquo;RE:Why&amp;rdquo; for extracting answers to a &amp;ldquo;Why&amp;rdquo; type question. RE:Why adopts new method which pays attention to the structual relations between sentences of a question and sentences of the answer on the Web. We have conducted experiments using twenty questions and have compared the performances between RE:Why and a traditional keyword search engine by measuring precisions and recalls.

DOI： 10.11309/jssst.24.3_20

J-GLOBAL

researchmap

Language：Japanese   Publisher：The Institute of Electronics, Information and Communication Engineers  

When we watch a movie which has captions, it is difficult to look an actor and a caption at the same time because the position of a caption is far from the face of an actor. We developed a system for placing a balloon near by an actor and displaying captions in the balloon. In order to detect an actor who is speaking at one scene, the system recognizes the two states of a mouth: open and close, where the position of the mouth region is estimated according to a result of face detection. In addition, the system decides an appropriate position for a balloon in order to avoid a situation such that a balloon overlaps the faces of other actors. In order to evaluate the efficiency of the proposal, we have provided a conversation scene of around .1. minute to 15 participants. And we compared the proposed method and a normal method, which is a method of displaying captions below the screen, using conversation scenes of around 1 minute. In this experiment, 15 participants watched these conversation scenes and answered a questionnaire. The results of the questionnaire have shown the proposed method is superior to the normal method at a point of easiness of watching a movie when a balloon is displayed at an appropriate position.

CiNii Books

J-GLOBAL

researchmap

Language：Japanese  

J-GLOBAL

researchmap

Language：Japanese  

J-GLOBAL

researchmap

Language：Japanese  

J-GLOBAL

researchmap

Language：Japanese   Publisher：The Institute of Electronics, Information and Communication Engineers  

In this paper, we propose a speech/music/noise discrimination technique using image features of sonagram (a 2-dimension image, the horizontal axis of which represents time, the vertical axis of which represents frequency, and the intensity of each point in which represents power of a particular frequency at a particular time). In the proposal technique, we focus on a difference in the sonagram between each audio section. In case of speech, gentle curves are drawn on the sonagram because the peak frequency of speech changes gently. In case of music, horizontal lines are drawn on the sonagram because the peak frequency of music is stable. And in case of noise, random points are drawn on the sonagram because the peak frequency of noise is unstable. Therefore, we calculate the 2-dimension frequency of the sonagram and discriminate three kinds of audio sections using discriminant analysis. We evaluated the proposal technique by precision and error rate of discrimination.

CiNii Books

J-GLOBAL

researchmap

Language：Japanese   Publisher：The Institute of Electronics, Information and Communication Engineers  

This paper proposes a method for detection of adult images. The method premises that intended adult images include lots of skin regions, the skin region are homogeneous and agglomerative. The method first makes skin likelihood images using GMM(Gaussian Mixture Model). The method second detects homogeneous skin regions in skin likelihood images. The method dynamically make a threshold every regions, then it's final skin regions that skin regions are grown using the threshold with regard to homogeneousness and agglomeration. And feature vectors are extracted from resized skin likelihood images of marked skin regions. Feature vectors are applied to PCA(principal component analysis) and ICA(independent component analysis). SVM are trained by feature vectors, then trained SVM classifies adult images or not.

CiNii Books

J-GLOBAL

researchmap

Language：English  

Contents of sports tickers in broadcasted sports videos could be efficient as queries in analyzing the video contents. However, since sports tickers are displayed with various colors and brightness, it is difficult to recognize inner objects of the sports tickers only according to the brightness distribution. Besides, in order to recognize what the objects represent, it is indispensable to investigate the change of these objects according to time traveling. In this paper, the proposed method can recognize locations of the out ticker, score tickers, and runner markers in the sports ticker of broadcasted baseball videos analyzing the brightness histograms obtained from every of grayscale and color images, and the change of the sports ticker with the correlation coefficient. Experimental results with four broadcasted baseball videos have shown that the proposed method can exactly work in the case when the out ticker is represented by a letter. © 2007, The Institute of Image Electronics Engineers of Japan. All rights reserved.

DOI： 10.11371/iieej.36.701

Scopus

J-GLOBAL

researchmap

Language：Japanese  

J-GLOBAL

researchmap

Language：Japanese  

J-GLOBAL

researchmap

Language：Japanese   Publisher：Information Processing Society of Japan (IPSJ)  

This paper reports on a method for figures-grounds discrimination for vector images based on contour-line extraction. The method is available for vector images such as clipart drawings. A vector image is expressed by composing plural objects (e.g. polygons, ellipses, lines, etc.), and some of the objects defines the region of the boundary between figure and ground. Assuming that the outermost contour-line in an image is the boundary between figure and ground, the method finds objects defining the region of the boundary. In contourline extraction, the method selects objects which have thin continuous regions and have high color-diffrence to their neiboruing regions. We have examined the performance of the method by use of 126 samples of vector images. In 81% of the samples, the method has automatically extracted figures.

CiNii Books

J-GLOBAL

researchmap

Language：Japanese   Publisher：Information Processing Society of Japan (IPSJ)  

Genetic Algorithms (GAs) are optimization methods inspired by evolutionary mechanisms. GAs fall into states which can't improve solutions (evolutionary stagnation) because indivisuals are always evoluated in order to improve. GAs have a difficulty problem to obtain the optimal solutions. As a way in order to dissolve this problem, the diversity maintenance of population, however, the problem occurs, which convergence speed becomes slow. This paper proposes a novel GA to avoid evolutionary stagnation by corrupting the solutions and also convergence speed does't become slow, by using improving population which improves the solutions, and corrupting population which corrupts the solutions, This paper also shows that good results compared with traditional GA on vehicle routing problems with time windows and function optimization problems.

CiNii Books

J-GLOBAL

researchmap

Language：Japanese   Publisher：Information Processing Society of Japan (IPSJ)  

The growth of Ubiquitous Computing makes the novel environment that provides context-aware services. Context-aware service applications have to discover sensors sensing proper context and having proper scopes, and get context data from the sensors. Therefore a new framework for decribing applications to grasp information of rooms and sensors is needed. One of ways to manage information of real world is using a world model constructed on computer system. This paper presents a world model construction system that provides 3D-shapes of room, location of sensors and sensor scopes for service applications. The system automatically constructs a world model acquiring and using information about rooms and sensors.

CiNii Books

J-GLOBAL

researchmap

Language：Japanese  

J-GLOBAL

researchmap

Language：Japanese  

J-GLOBAL

researchmap

Language：Japanese  

J-GLOBAL

researchmap

Language：Japanese  

J-GLOBAL

researchmap

Language：Japanese   Publisher：The Institute of Electronics, Information and Communication Engineers  

This paper reports a new method for retrieval of landscape images using query by sketch and semantic icon. In order to retrieve images from a database, a user sketches shapes of landscape objects (e.g. mountain, river, road, etc.) and places landscape icons on the sketch. In the retrieval process, our method examines for each image in the database whether an image matches the user-sketch of landscape objects. We have evaluated the efficiency of the retrieval method and have obtained the result that 80 % images for the top 10 hits of the retrieval results are matched to the user-sketched patterns.

CiNii Books

J-GLOBAL

researchmap

Language：Japanese  

J-GLOBAL

researchmap

Language：Japanese  

J-GLOBAL

researchmap

Language：Japanese  

J-GLOBAL

researchmap

Language：Japanese  

J-GLOBAL

researchmap

Language：Japanese  

J-GLOBAL

researchmap

Language：Japanese   Publisher：Japan Society for Software Science and Technology  

We propose a system MineBlog for arousing new interests by recommending blog articles. We assume that articles arousing user's interest involve both similar and different viewpoints of the user. MineBlog processes as follows. First, it gathers blog articles from the web, and stores them into a database. Second, it extracts feature terms from the article written by the user and the articles stored in the database, it selects articles which contain the same feature terms and regards them as candidates for recommendation. It scores them using relevance and rarity measures, which are quantified degrees of similarity and difference between the user's and other's viewpoints, and recommends the user high-scored ones. We have implemented MineBlog and evaluated the effectiveness by experiments.

DOI： 10.11309/jssstconference.21.0.76.0

J-GLOBAL

researchmap

Language：Japanese  

J-GLOBAL

researchmap

Language：Japanese  

J-GLOBAL

researchmap

Language：Japanese  

J-GLOBAL

researchmap

Language：Japanese  

J-GLOBAL

researchmap

Language：Japanese  

J-GLOBAL

researchmap

Language：Japanese  

J-GLOBAL

researchmap