IEICE Transactions on Information and Systems
Online ISSN : 1745-1361
Print ISSN : 0916-8532
E93.D 巻, 4 号
選択された号の論文の27件中1~27を表示しています
Special Section on Knowledge-Based Software Engineering
  • Atsushi OHNISHI
    2010 年 E93.D 巻 4 号 p. 677-678
    発行日: 2010/04/01
    公開日: 2010/04/01
    ジャーナル フリー
  • Noboru HATTORI, Shuichiro YAMAMOTO, Tsuneo AJISAKA, Tsuyoshi KITANI
    原稿種別: PAPER
    専門分野: Requirements Engineering
    2010 年 E93.D 巻 4 号 p. 679-692
    発行日: 2010/04/01
    公開日: 2010/04/01
    ジャーナル フリー
    We propose requirement validation criteria and a method based on the interaction between actors in an information system. We focus on the cyclical transitions of one actor's situation against another and clarify observable stimuli and responses based on these transitions. Both actors' situations can be listed in a state transition table, which describes the observable stimuli or responses they send or receive. Examination of the interaction between both actors in the state transition tables enables us to detect missing or defective observable stimuli or responses. Typically, this method can be applied to the examination of the interaction between a resource managed by the information system and its user. As a case study, we analyzed 332 requirement defect reports of an actual system development project in Japan. We found that there were a certain amount of defects regarding missing or defective stimuli and responses, which can be detected using our proposed method if this method is used in the requirement definition phase. This means that we can reach a more complete requirement definition with our proposed method.
  • Atsushi OHNISHI, Koji KITAMOTO
    原稿種別: PAPER
    専門分野: Requirements Engineering
    2010 年 E93.D 巻 4 号 p. 693-701
    発行日: 2010/04/01
    公開日: 2010/04/01
    ジャーナル フリー
    This paper proposes a method to generate alternative scenarios from a normal scenario written with a scenario language. This method includes (1) generation of alternative plans and (2) generation of alternative scenario by a user's selection of these plans. The proposed method enables users to decrease the omission of the possible alternative scenarios in the early stages of development. The method will be illustrated with some examples.
  • Haruhiko KAIYA, Masaaki TANIGAWA, Shunichi SUZUKI, Tomonori SATO, Akir ...
    原稿種別: PAPER
    専門分野: Requirements Engineering
    2010 年 E93.D 巻 4 号 p. 702-712
    発行日: 2010/04/01
    公開日: 2010/04/01
    ジャーナル フリー
    Quality requirements are scattered over a requirements specification, thus it is hard to measure and trace such quality requirements to validate the specification against stakeholders' needs. We proposed a technique called “spectrum analysis for quality requirements” which enabled analysts to sort a requirements specification to measure and track quality requirements in the specification. In the same way as a spectrum in optics, a quality spectrum of a specification shows a quantitative feature of the specification with respect to quality. Therefore, we can compare a specification of a system to another one with respect to quality. As a result, we can validate such a specification because we can check whether the specification has common quality features and know its specific features against specifications of existing similar systems. However, our first spectrum analysis for quality requirements required a lot of effort and knowledge of a problem domain and it was hard to reuse such knowledge to reduce the effort. We thus introduce domain knowledge called term-characteristic map (TCM) to reuse the knowledge for our quality spectrum analysis. Through several experiments, we evaluate our spectrum analysis, and main finding are as follows. First, we confirmed specifications of similar systems have similar quality spectra. Second, results of spectrum analysis using TCM are objective, i.e., different analysts can generate almost the same spectra when they analyze the same specification.
  • Takehiko MURAKAWA, Masaru NAKAGAWA
    原稿種別: PAPER
    専門分野: Software Development Techniques
    2010 年 E93.D 巻 4 号 p. 713-720
    発行日: 2010/04/01
    公開日: 2010/04/01
    ジャーナル フリー
    Thinking process development diagram is a graphical expression from which readers can easily find not only the hierarchy of a given problem but the relationship between the problem and the solution. Although that has been developed as an idea creation support tool in the field of mechanical design, we referred to the restricted version as clamshell diagram to attempt to apply to other fields. In this paper we propose the framework for drawing the diagram of the SQL statement. The basic idea is to supply the hierarchical code fragments of a given SQL statement in the left side of the diagram and to put the meaning written in a natural language in the right. To verify the usefulness of the diagram expression, we actually drew several clamshell diagrams. For three SQL statements that are derived from the same specification, the resulting diagrams enable us to understand the difference visually.
  • Junko SHIROGANE, Takashi MORI, Hajime IWATA, Yoshiaki FUKAZAWA
    原稿種別: PAPER
    専門分野: Software Development Techniques
    2010 年 E93.D 巻 4 号 p. 721-732
    発行日: 2010/04/01
    公開日: 2010/04/01
    ジャーナル フリー
    Computers and software have become necessities in human society, and most people are required to be able to use them. However, software is not always accessible for users with special needs, and it is difficult to develop software accessible to such users. There are many guidelines and support tools for developing accessible websites. For software, however, such guidelines and support tools are few. In our research, to develop accessible software easily we propose a method of evaluating the accessibility of Graphical User Interface (GUI) software. This method involves analyzing GUI software source programs, assessing accessibility on the basis of accessibility guidelines, and presenting a list of unsatisfactory accessibility code and indicating how to modify it.
  • Teruyoshi ZENMYO, Takashi KOBAYASHI, Motoshi SAEKI
    原稿種別: PAPER
    専門分野: Software Development Techniques
    2010 年 E93.D 巻 4 号 p. 733-744
    発行日: 2010/04/01
    公開日: 2010/04/01
    ジャーナル フリー
    One of the critical issue in framework-based software development is a huge introduction cost caused by technical gap between developers and users of frameworks. This paper proposes a technique for deriving framework usages to implement a given requirements specification. By using the derived usages, the users can use the frameworks without understanding the framework in detail. Requirements specifications which describe definite behavioral requirements cannot be related to frameworks in as-is since the frameworks do not have definite control structure so that the users can customize them to suit given requirements specifications. To cope with this issue, a new technique based on satisfiability problems (SAT) is employed to derive the control structures of the framework model. In the proposed technique, requirements specifications and frameworks are modeled based on Labeled Transition Systems (LTSs) with branch conditions represented by predicates. Truth assignments of the branch conditions in the framework models are not given initially for representing the customizable control structure. The derivation of truth assignments of the branch conditions is regarded as the SAT by assuming relations between termination states of the requirements specification model and ones of the framework model. This derivation technique is incorporated into a technique we have proposed previously for relating actions of requirements specifications to ones of frameworks. Furthermore, this paper discuss a case study of typical use cases in e-commerce systems.
  • Shozo HORI, Takako NAKATANI, Keiichi KATAMINE, Naoyasu UBAYASHI, Masaa ...
    原稿種別: PAPER
    専門分野: Management Techniques
    2010 年 E93.D 巻 4 号 p. 745-753
    発行日: 2010/04/01
    公開日: 2010/04/01
    ジャーナル フリー
    We propose PM (Project Management) patterns to prevent schedule delays caused by changes in requirements on empirical studies. Changes or late elicitation of requirements during the design, coding and test processes are one of the most serious risks, which may delay project schedules. However, changes and late elicitation of requirements are usually accepted during development processes. Therefore, the PM methods for preventing schedule delays caused by changes and late elicitation of requirements during development processes are an important area of study. In this study, we examined the actual conditions of various projects which succeeded in preventing schedule delays resulting from changes and late elicitation of requirements during development processes. We were able to extract various typical PM techniques for preventing these schedule delays. The techniques, known as “PM patterns”, were also applied to other projects. The patterns were arranged on a two-dimensional framework. We discuss a framework of PM patterns aimed at solving the problems caused by changes in requirements.
  • Shinpei HAYASHI, Yasuyuki TSUDA, Motoshi SAEKI
    原稿種別: PAPER
    専門分野: Management Techniques
    2010 年 E93.D 巻 4 号 p. 754-762
    発行日: 2010/04/01
    公開日: 2010/04/01
    ジャーナル フリー
    This paper proposes a technique for detecting the occurrences of refactoring from source code revisions. In a real software development process, a refactoring operation may sometimes be performed together with other modifications at the same revision. This means that detecting refactorings from the differences between two versions stored in a software version archive is not usually an easy process. In order to detect these impure refactorings, we model the detection within a graph search. Our technique considers a version of a program as a state and a refactoring as a transition between two states. It then searches for the path that approaches from the initial to the final state. To improve the efficiency of the search, we use the source code differences between the current and the final state for choosing the candidates of refactoring to be applied next and estimating the heuristic distance to the final state. Through case studies, we show that our approach is feasible to detect combinations of refactorings.
  • Masao OKABE, Akiko YOSHIOKA, Keido KOBAYASHI, Takahira YAMAGUCHI
    原稿種別: PAPER
    専門分野: Management Techniques
    2010 年 E93.D 巻 4 号 p. 763-773
    発行日: 2010/04/01
    公開日: 2010/04/01
    ジャーナル フリー
    In recent automated and integrated manufacturing, so-called intelligence skill is becoming more and more important and its efficient transfer to next-generation engineers is one of the urgent issues. In this paper, we propose a new approach without costly OJT (on-the-job training), that is, combinational usage of a domain ontology, a rule ontology and a rule-based system. Intelligence skill can be decomposed into pieces of simple engineering rules. A rule ontology consists of these engineering rules as primitives and the semantic relations among them. A domain ontology consists of technical terms in the engineering rules and the semantic relations among them. A rule ontology helps novices get the total picture of the intelligence skill and a domain ontology helps them understand the exact meanings of the engineering rules. A rule-based system helps domain experts externalize their tacit intelligence skill to ontologies and also helps novices internalize them. As a case study, we applied our proposal to some actual job at a remote control and maintenance office of hydroelectric power stations in Tokyo Electric Power Co., Inc. We also did an evaluation experiment for this case study and the result supports our proposal.
  • Sameera ABAR, Tetsuo KINOSHITA
    原稿種別: PAPER
    専門分野: Management Techniques
    2010 年 E93.D 巻 4 号 p. 774-788
    発行日: 2010/04/01
    公開日: 2010/04/01
    ジャーナル フリー
    This paper presents a domain-ontology driven multi-agent based scheme for representing the knowledge of the communication network management system. In the proposed knowledge-intensive framework, the static domain-related concepts are articulated as the domain knowledge ontology. The experiential knowledge for managing the network is represented as the fault-case reasoning models, and it is explicitly encoded as the core knowledge of multi-agent middleware layer as heuristic production-type rules. These task-oriented management expertise manipulates the domain content and structure during the diagnostic sessions. The agents' rules along with the embedded generic java-based problem-solving algorithms and run-time log information, perform the automated management tasks. For the proof of concept, an experimental network system has been implemented in our laboratory, and the deployment of some test-bed scenarios is performed. Experimental results confirm a marked reduction in the management-overhead of the network administrator, as compared to the manual network management techniques, in terms of the time-taken and effort-done during a particular fault-diagnosis session. Validation of the reusability/modifiability aspects of our system, illustrates the flexible manipulation of the knowledge fragments within diverse application contexts. The proposed approach can be regarded as one of the pioneered steps towards representing the network knowledge via reusable domain ontology and intelligent agents for the automated network management support systems.
  • Ichiro SATOH
    原稿種別: PAPER
    専門分野: Agent-based System
    2010 年 E93.D 巻 4 号 p. 789-799
    発行日: 2010/04/01
    公開日: 2010/04/01
    ジャーナル フリー
    This paper presents an agent-based system for building and operating context-aware services in public spaces, including museums. The system provides users with agents and detects the locations of users and deploys location-aware user-assistant agents at computers near the their current locations by using active RFID-tags. When a visitor moves between exhibits in a museum, this dynamically deploys his/her agent at the computers close to the exhibits by using mobile agent technology. It annotates the exhibits in his/her personalized form and navigate him/her user to the next exhibits along his/her routes. It also introduces user movement as a natural approach to interacting between users and agents. To demonstrate the utility and effectiveness of the system, we constructed location/user-aware visitor-guide services and experimented them for two weeks in a public museum.
Regular Section
  • Shigeto TAJIMA, Nobuo FUNABIKI, Teruo HIGASHINO
    原稿種別: PAPER
    専門分野: Fundamentals of Information Systems
    2010 年 E93.D 巻 4 号 p. 800-810
    発行日: 2010/04/01
    公開日: 2010/04/01
    ジャーナル フリー
    Wireless mesh networks have been extensively studied as expandable, flexible, and inexpensive access networks to the Internet. This paper focuses on one composed of multiple access points (APs) connected through multihop wireless communications mainly by the wireless distribution system (WDS). For scalability, the proper partition of APs into multiple WDS clusters is essential, because the number of APs in one cluster is limited due to the increasing radio interference and control packets. In this paper, we formulate this WDS clustering problem and prove the NP-completeness of its decision version through reduction from a known NP-complete problem. Then, we propose its heuristic algorithm, using a greedy method and a variable depth search method, to satisfy the complex constraints while optimizing the cost function. We verify the effectiveness of our algorithm through extensive simulations, where the results confirm its superiority to the existing algorithm in terms of throughput.
  • Shuang ZHAO, Wenqing LU, Xiaofang ZHOU, Dian ZHOU, Gerald E. SOBELMAN
    原稿種別: PAPER
    専門分野: Fundamentals of Information Systems
    2010 年 E93.D 巻 4 号 p. 811-821
    発行日: 2010/04/01
    公開日: 2010/04/01
    ジャーナル フリー
    MIMO-OFDM systems aim to improve transmission quality and/or throughput but require significant signal processing capability and flexibility at reasonable cost. This paper proposes a reconfigurable architecture and associated algorithm optimizations for these types of systems based on the IEEE 802.11n and IEEE 802.16e standards. In particular, we describe the implementation of two key computations onto this architecture, namely Fast Fourier Transform (FFT) and Space-Time Block Decoding (STBD). The design is post-layout using a UMC 0.18micron technology at a clock rate of 100MHz. Performance comparisons with other optimization methods and hardware implementations are given.
  • Hiroshi IGAKI, Masahide NAKAMURA
    原稿種別: PAPER
    専門分野: Software System
    2010 年 E93.D 巻 4 号 p. 822-833
    発行日: 2010/04/01
    公開日: 2010/04/01
    ジャーナル フリー
    This paper presents a framework for formalizing and detecting feature interactions (FIs) in the emerging smart home domain. We first establish a model of home network system (HNS), where every networked appliance (or the HNS environment) is characterized as an object consisting of properties and methods. Then, every HNS service is defined as a sequence of method invocations of the appliances. Within the model, we next formalize two kinds of FIs: (a) appliance interactions and (b) environment interactions. An appliance interaction occurs when two method invocations conflict on the same appliance, whereas an environment interaction arises when two method invocations conflict indirectly via the environment. Finally, we propose offline and online methods that detect FIs before service deployment and during execution, respectively. Through a case study with seven practical services, it is shown that the proposed framework is generic enough to capture feature interactions in HNS integrated services. We also discuss several FI resolution schemes within the proposed framework.
  • Yuna KIM, Wan Yeon LEE, Kyong Hoon KIM, Jong KIM
    原稿種別: PAPER
    専門分野: Data Engineering, Web Information Systems
    2010 年 E93.D 巻 4 号 p. 834-842
    発行日: 2010/04/01
    公開日: 2010/04/01
    ジャーナル フリー
    In this paper, we propose a novel Web service composition framework which dynamically accommodates various failure recovery requirements. In the proposed framework called Adaptive Failure-handling Framework (AdaFF), failure-handling submodules are prepared during the design of a composite service, and some of them are systematically selected and automatically combined with the composite Web service at service instantiation in accordance with the requirement of individual users. In contrast, existing frameworks cannot adapt the failure-handling behaviors to user's requirements. AdaFF rapidly delivers a composite service supporting the requirement-matched failure handling without manual development, and contributes to a flexible composite Web service design in that service architects never care about failure handling or variable requirements of users. For proof of concept, we implement a prototype system of the AdaFF, which automatically generates a composite service instance with Web Services Business Process Execution Language (WS-BPEL) according to the users' requirement specified in XML format and executes the generated instance on the ActiveBPEL engine.
  • Lihong MA, Dong YU, Gang WEI, Jing TIAN, Hanqing LU
    原稿種別: PAPER
    専門分野: Information Network
    2010 年 E93.D 巻 4 号 p. 843-857
    発行日: 2010/04/01
    公開日: 2010/04/01
    ジャーナル フリー
    Major challenges of the conventional spread-transform dither modulation (STDM) watermarking approach are two-fold: (i) it exploits a fixed watermarking strength (more particularly, the quantization index step size) to the whole cover image; and (ii) it is fairly vulnerable to the amplitude changes. To tackle the above challenges, an adaptive spread-transform dither modulation (ASTDM) approach is proposed in this paper for conducting robust color image watermarking by incorporating a new perceptual model into the conventional STDM framework. The proposed approach exploits a new perceptual model to adjust the quantization index step sizes according to the local perceptual characteristics of a cover image. Furthermore, in contrast to the conventional Watson's model is vulnerable to the amplitude changes, our proposed new perceptual model makes the luminance masking thresholds be consistent with any amplitude change, while keeping the consistence to the properties of the human visual system. In addition, certain color artifacts could be incurred during the watermark embedding procedure, since some intensity values are perceptibly changed to label the watermark. For that, a color artifact suppression algorithm is proposed by mathematically deriving an upper bound for the intensity values according to the inherent relationship between the saturation and the intensity components. Extensive experiments are conducted using 500 images selected from Corel database to demonstrate the superior performance of the proposed ASTDM approach.
  • Xianzhi YE, Lei JING, Mizuo KANSEN, Junbo WANG, Kaoru OTA, Zixue CHENG
    原稿種別: PAPER
    専門分野: Educational Technology
    2010 年 E93.D 巻 4 号 p. 858-872
    発行日: 2010/04/01
    公開日: 2010/04/01
    ジャーナル フリー
    With the progress of ubiquitous technology, ubiquitous learning presents new opportunities to learners. Situations of a learner can be grasped through analyzing the learner's actions collected by sensors, RF-IDs, or cameras in order to provide support at proper time, proper place, and proper situation. Training for acquiring skills and enhancing physical abilities through exercise and experience in the real world is an important domain in u-learning. A training program may last for several days and has one or more training units (exercises) for a day. A learner's performance in a unit is considered as short term state. The performance in a series of units may change with patterns: progress, plateau, and decline. Long term state in a series of units is accumulatively computed based on short term states. In a learning/training program, it is necessary to apply different support strategies to adapt to different states of the learner. Adaptation in learning support is significant, because a learner loses his/her interests easily without adaptation. Systems with the adaptive support usually provide stimulators to a learner, and a learner can have a great motivation in learning at beginning. However, when the stimulators reach some levels, the learner may lose his/her motivation, because the long term state of the learner changes dynamically, which means a progress state may change to a plateau state or a decline state. In different long term learning states, different types of stimulators are needed. However, the stimulators and advice provided by the existing systems are monotonic without changeable support strategies. We propose a mutual adaptive support. The mutual adaptation means each of the system and the learner has their own states. On one hand, the system tries to change its state to adapt to the learner's state for providing adaptive support. On the other hand, the learner can change its performance following the advice given based on the state of the system. We create a ubiquitous pet (u-pet) as a metaphor of our system. A u-pet is always with the learner and encourage the leaner to start training at proper time and to do training smoothly. The u-pet can perform actions with the learner in training, change its own attributes based on the learner's attributes, and adjust its own learning rate by a learning function. The u-pet grasps the state of the learner and adopts different training support strategies to the learner's training based on the learner's short and long term states.
  • Chung-Lin WEN, Bing-Yu CHEN, Yoichi SATO
    原稿種別: PAPER
    専門分野: Image Processing and Video Processing
    2010 年 E93.D 巻 4 号 p. 873-881
    発行日: 2010/04/01
    公開日: 2010/04/01
    ジャーナル フリー
    In this paper, we present an interactive and intuitive graph-cut-based video segmentation system while taking both color and motion information into consideration with a stroke-based user interface. Recently, graph-cut-based methods become prevalent for image and video segmentation. However, most of them deal with color information only and usually failed under circumstances where there are some regions in both foreground and background with similar colors. Unfortunately, it is usually hard to avoid, especially when the objects are filmed under a natural environment. To make such methods more practical to use, we propose a graph-cut-based video segmentation method based on both color and motion information, since the foreground objects and the background usually have different motion patterns. Moreover, to make the refinement mechanism easy to use, the strokes drawn by the user are propagated to the temporal-spatial video volume according to the motion information for visualization, so that the user can draw some additional strokes to refine the segmentation result in the video volume. The experiment results show that by combining both color and motion information, our system can resolve the wrong labeling due to the color similarity, even the foreground moving object is behind an occlusion object.
  • Ahmed AFIFI, Toshiya NAKAGUCHI, Norimichi TSUMURA, Yoichi MIYAKE
    原稿種別: PAPER
    専門分野: Image Recognition, Computer Vision
    2010 年 E93.D 巻 4 号 p. 882-890
    発行日: 2010/04/01
    公開日: 2010/04/01
    ジャーナル フリー
    The aim of this work is to develop an efficient medical image segmentation technique by fitting a nonlinear shape model with pre-segmented images. In this technique, the kernel principle component analysis (KPCA) is used to capture the shape variations and to build the nonlinear shape model. The pre-segmentation is carried out by classifying the image pixels according to the high level texture features extracted using the over-complete wavelet packet decomposition. Additionally, the model fitting is completed using the particle swarm optimization technique (PSO) to adapt the model parameters. The proposed technique is fully automated, is talented to deal with complex shape variations, can efficiently optimize the model to fit the new cases, and is robust to noise and occlusion. In this paper, we demonstrate the proposed technique by implementing it to the liver segmentation from computed tomography (CT) scans and the obtained results are very hopeful.
  • Ken-ichi SUZUKI, Yoshiyuki KAERIYAMA, Kazuhiko KOMATSU, Ryusuke EGAWA, ...
    原稿種別: PAPER
    専門分野: Computer Graphics
    2010 年 E93.D 巻 4 号 p. 891-902
    発行日: 2010/04/01
    公開日: 2010/04/01
    ジャーナル フリー
    Ray tracing is one of the most popular techniques for generating photo-realistic images. Extensive research and development work has made interactive static scene rendering realistic. This paper deals with interactive dynamic scene rendering in which not only the eye point but also the objects in the scene change their 3D locations every frame. In order to realize interactive dynamic scene rendering, RTRPS (Ray Tracing based on Ray Plane and Bounding Sphere), which utilizes the coherency in rays, objects, and grouped-rays, is introduced. RTRPS uses bounding spheres as the spatial data structure which utilizes the coherency in objects. By using bounding spheres, RTRPS can ignore the rotation of moving objects within a sphere, and shorten the update time between frames. RTRPS utilizes the coherency in rays by merging rays into a ray-plane, assuming that the secondary rays and shadow rays are shot through an aligned grid. Since a pair of ray-planes shares an original ray, the intersection for the ray can be completed using the coherency in the ray-planes. Because of the three kinds of coherency, RTRPS can significantly reduce the number of intersection tests for ray tracing. Further acceleration techniques for ray-plane-sphere and ray-triangle intersection are also presented. A parallel projection technique converts a 3D vector inner product operation into a 2D operation and reduces the number of floating point operations. Techniques based on frustum culling and binary-tree structured ray-planes optimize the order of intersection tests between ray-planes and a sphere, resulting in 50% to 90% reduction of intersection tests. Two ray-triangle intersection techniques are also introduced, which are effective when a large number of rays are packed into a ray-plane. Our performance evaluations indicate that RTRPS gives 13 to 392 times speed up in comparison with a ray tracing algorithm without organized rays and spheres. We found out that RTRPS also provides competitive performance even if only primary rays are used.
  • Ayako NISHIMURA, Minoru SAKAIRI, Daisuke SUZUKI
    原稿種別: PAPER
    専門分野: Multimedia Pattern Processing
    2010 年 E93.D 巻 4 号 p. 903-908
    発行日: 2010/04/01
    公開日: 2010/04/01
    ジャーナル フリー
    We have developed an odor-emitting apparatus for application of odor to information technology. This apparatus consists of a chemical capsule cartridge including chemical capsules of odor ingredients and valves to control odor emission using an artificial metal muscle. In this method, multiple valves can be opened using the current for a single artificial muscle because the expansion and contraction time constant for the artificial muscles is large. We have developed a new multi-valve sequence mode that uses multiple odor capsules to increase odor strength, and we have been able to increase the strength produced by a factor of two. In addition, we evaluated the change in odor strength using a mock-up of the back seat of an automobile, and all of the ten test subjects reported sensing a stronger odor.
  • Ithipan METHASATE, Thanaruk THEERAMUNKONG
    原稿種別: PAPER
    専門分野: Biocybernetics, Neurocomputing
    2010 年 E93.D 巻 4 号 p. 909-921
    発行日: 2010/04/01
    公開日: 2010/04/01
    ジャーナル フリー
    Finding a kernel mapping function for support vector machines (SVMs) is a key step towards construction of a high-performanced SVM-based classifier. While some recent methods exploited an evolutional approach to construct a suitable multifunction kernel, most of them searched randomly and diversely. In this paper, the concept of a family of identical-structured kernel trees is proposed to enable exploration of structure space using genetic programming whereas to pursue investigation of parameter space on a certain tree using evolution strategy. To control balance between structure and parameter search towards an optimal kernel, simulated annealing is introduced. By experiments on a number of benchmark datasets in the UCI and text classification collection, the proposed method is shown to be able to find a better optimal solution than other search methods, including grid search and gradient search.
  • Yu Gwang JIN, Nam Soo KIM
    原稿種別: LETTER
    専門分野: Pattern Recognition
    2010 年 E93.D 巻 4 号 p. 922-925
    発行日: 2010/04/01
    公開日: 2010/04/01
    ジャーナル フリー
    In this paper, we propose a novel target acoustic signal detection approach which is based on non-negative matrix factorization (NMF). Target basis vectors are trained from the target signal database through NMF, and input vectors are projected onto the subspace spanned by these target basis vectors. By analyzing the distribution of time-varying normalized projection error, the optimal threshold can be calculated to detect the target signal intervals during the entire input signal. Experimental results show that the proposed algorithm can detect the target signal successfully under various signal environments.
  • Jik-Han JUNG, Hwal-Suk LEE, Dong-Jo PARK
    原稿種別: LETTER
    専門分野: Image Processing and Video Processing
    2010 年 E93.D 巻 4 号 p. 926-929
    発行日: 2010/04/01
    公開日: 2010/04/01
    ジャーナル フリー
    In this letter, a novel technique for fast block matching using a new matching criterion is proposed. The matching speed and image quality are controlled by the one control parameter called matching region ratio. An efficient matching scheme with a gradual voting strategy is also proposed. This scheme can greatly boost the matching speed. The proposed technique is fast and applicable even in the presence of speckle noise or partial occlusion.
  • Wenjie XIE, De XU, Shuoyan LIU, Yingjun TANG
    原稿種別: LETTER
    専門分野: Image Recognition, Computer Vision
    2010 年 E93.D 巻 4 号 p. 930-933
    発行日: 2010/04/01
    公開日: 2010/04/01
    ジャーナル フリー
    This paper focuses on the relationship between the number of interest points and the accuracy rate in scene classification. Here, we accept the common belief that more interest points can generate higher accuracy. But, few effort have been done in this field. In order to validate this viewpoint, in our paper, extensive experiments based on bag of words method are implemented. In particular, three different SIFT descriptors and five feature selection methods are adopted to change the number of interest points. As innovation point, we propose a novel dense SIFT descriptor named Octave Dense SIFT, which can generate more interest points and higher accuracy, and a new feature selection method called number mutual information (NMI), which has better robustness than other feature selection methods. Experimental results show that the number of interest points can aggressively affect classification accuracy.
  • Sheng LI, Xiao-Yuan JING, Lu-Sha BIAN, Shi-Qiang GAO, Qian LIU, Yong-F ...
    原稿種別: LETTER
    専門分野: Image Recognition, Computer Vision
    2010 年 E93.D 巻 4 号 p. 934-937
    発行日: 2010/04/01
    公開日: 2010/04/01
    ジャーナル フリー
    In this letter, a statistical uncorrelated near class discriminant (SUNCD) approach is proposed for face recognition. The optimal discriminant vector obtained by this approach can differentiate one class and its near classes, i.e., its nearest neighbor classes, by constructing the specific between-class and within-class scatter matrices and using the Fisher criterion. In this manner, SUNCD acquires all discriminant vectors class by class. Furthermore, SUNCD makes every discriminant vector satisfy locally statistical uncorrelated constraints by using the corresponding class and part of its most neighboring classes. Experiments on the public AR face database demonstrate that the proposed approach outperforms several representative discriminant methods.
feedback
Top