Lazaro S.P. BUSAGALA Wataru OHYAMA Tetsushi WAKABAYASHI Fumitaka KIMURA
Feature transformation in automatic text classification (ATC) can lead to better classification performance. Furthermore dimensionality reduction is important in ATC. Hence, feature transformation and dimensionality reduction are performed to obtain lower computational costs with improved classification performance. However, feature transformation and dimension reduction techniques have been conventionally considered in isolation. In such cases classification performance can be lower than when integrated. Therefore, we propose an integrated feature analysis approach which improves the classification performance at lower dimensionality. Moreover, we propose a multiple feature integration technique which also improves classification effectiveness.
Donghyuk SHIN Jeongseok HA Kyoungwoo HEO Hyuckjae LEE
We propose a new stopping criterion for decoding LDPC codes which consists of a measure of decoder behaviors and a decision rule to predict decoding failure. We will show that the proposed measure, the number of satisfied check nodes, does not need (or minimizes) additional complexity, and the decision rule is efficient and more importantly channel independent, which was not possible in the previous work.
Hongbin SUO Ming LI Ping LU Yonghong YAN
Robust automatic language identification (LID) is the task of identifying the language from a short utterance spoken by an unknown speaker. The mainstream approaches include parallel phone recognition language modeling (PPRLM), support vector machine (SVM) and the general Gaussian mixture models (GMMs). These systems map the cepstral features of spoken utterances into high level scores by classifiers. In this paper, in order to increase the dimension of the score vector and alleviate the inter-speaker variability within the same language, multiple data groups based on supervised speaker clustering are employed to generate the discriminative language characterization score vectors (DLCSV). The back-end SVM classifiers are used to model the probability distribution of each target language in the DLCSV space. Finally, the output scores of back-end classifiers are calibrated by a pair-wise posterior probability estimation (PPPE) algorithm. The proposed language identification frameworks are evaluated on 2003 NIST Language Recognition Evaluation (LRE) databases and the experiments show that the system described in this paper produces comparable results to the existing systems. Especially, the SVM framework achieves an equal error rate (EER) of 4.0% in the 30-second task and outperforms the state-of-art systems by more than 30% relative error reduction. Besides, the performances of proposed PPRLM and GMMs algorithms achieve an EER of 5.1% and 5.0% respectively.
In this paper, an automatic identification method based on frequency discrimination is proposed. The proposed method can be used when the received signal is a constant envelope modulation scheme. To test the proposed method PSK and FSK are considered. Using computer simulations, the performance of the proposed method was evaluated and found to be able to distinguish between PSK and FSK well even in the presence of noise.
Video multicast over wireless medium has gained increasing popularity in a wide range of applications, such as video-on-demand and group video conferencing. With mobile ad hoc networks emerging as a promising solution for future ubiquitous communications, supporting reliable video multicast over mobile ad hoc networks is a timely research topic. In this paper we tackle this issue by using multiple tree multicast routing protocol. Specifically, we introduce an extension to the Multicast Ad Hoc On-demand Distance Vector (MAODV) routing protocol to construct two optimally disjoint trees in a single routine. The extended protocol is called Multiple Tree Multicast Ad Hoc On-demand Distance Vector (MT-MAODV) routing protocol. In order to distribute the video evenly and independently between these disjoint trees, the Multiple Description Coding (MDC) scheme is used for video coding. Simulation shows that the proposed protocol demonstrates video multicast with better quality than the conventional video multicast using a single tree only.
Landobasa Y.M.A.L. TOBING Pieter DUMON Roel BAETS Desmond. C.S. LIM Mee-Koy CHIN
We propose and demonstrate a simple one-bus two-ring configuration where the two rings are mutually coupled that has advantages over the one-ring structure. Unlike a one cavity system, it can exhibit near critically-coupled transmission with a broader range of loss. It can also significantly enhance the cavity finesse by simply making the second ring twice the size of the bus-coupled one, with the enhancement proportional to the intensity buildup in the second ring.
Due to the decoding procedure and filtering for edge detection, the feature extraction process of MPEG-7 Edge Histogram Descriptor (EHD) is time-consuming and computationally expensive. We proposed the fast EHD generation method in wavelet domain of JPEG2000 images. Experimental results demonstrate the advantage of this method over EHD.
Kazuya TSUKAMOTO Takeshi YAMAGUCHI Shigeru KASHIHARA Yuji OIE
In ubiquitous networks, Mobile Nodes (MNs) often suffer from performance degradation due to the following two reasons: (1) reduction of signal strength by the movement of an MN and intervening objects, and (2) radio interference with other WLANs. Therefore, handover initiation based on quick and reliable detection of the deterioration in a wireless link condition arising from the above two reasons is essential for achieving seamless handover. In previous studies, we focused on a handover decision criterion and described the problems related to the two existing decision criteria. Furthermore, we showed the effectiveness of the number of frame retransmissions through simulation experiments. However, a comparison of the signal strength and the number of frame retransmissions could not be examined due to the unreliability of the signal strength in simulations. Therefore, in the present paper, by employing FTP and VoIP applications, we compare the signal strength and the number of frame retransmissions as a handover decision criterion with experiments in terms of (1) and (2) in a real environment. Finally, we clarify the problem of the signal strength in contrast to the effectiveness of the number of frame retransmissions as a handover decision criterion.
Miyuki IMADA Masakatsu OHTA Mitsuo TERAMOTO Masayasu YAMAGUCHI
In this paper, we propose a method of controlling personal data disclosure based on LooM (Loosely Managed Privacy Protection Method) that prevents a malicious third party from identifying a person when he/she gets context-aware services using personal data. The basic function of LooM quantitatively evaluates the anonymity level of a person who discloses his/her data, and controls the personal-data disclosure according to the level. LooM uses a normalized entropy value for quantifying the anonymity. In this version of the LooM, the disclosure control is accomplished by adding two new functions. One is an abstracting-function that generates abstractions (or summaries) from the raw personal data to reduce the danger that the malicious third party might identify the person who discloses his/her personal data to the party. The other function is a unique-value-masking function that hides the unique personal data in the database. These functions enhance the disclosure control mechanism of LooM. We evaluate the functions using simulation data and questionnaire data. Then, we confirm the effectiveness of the functions. Finally, we show a prototype of a crime-information-sharing service to confirm the feasibility of these functions.
Yasushi HIDAKA Masashi SUGIYAMA
In order to obtain better generalization performance in supervised learning, model parameters should be determined appropriately, i.e., they should be determined so that the generalization error is minimized. However, since the generalization error is inaccessible in practice, the model parameters are usually determined so that an estimator of the generalization error is minimized. The regularized subspace information criterion (RSIC) is such a generalization error estimator for model selection. RSIC includes an additional regularization parameter and it should be determined appropriately for better model selection. A meta-criterion for determining the regularization parameter has also been proposed and shown to be useful in practice. In this paper, we show that there are several drawbacks in the existing meta-criterion and give an alternative meta-criterion that can solve the problems. Through simulations, we show that the use of the new meta-criterion further improves the model selection performance.
Shun GOKITA Masashi SUGIYAMA Keisuke SAKURAI
In order to obtain better learning results in supervised learning, it is important to choose model parameters appropriately. Model selection is usually carried out by preparing a finite set of model candidates, estimating a generalization error for each candidate, and choosing the best one from the candidates. If the number of candidates is increased in this procedure, the optimization quality may be improved. However, this in turn increases the computational cost. In this paper, we focus on a generalization error estimator called the regularized subspace information criterion and derive an analytic form of the optimal model parameter over a set of infinitely many model candidates. This allows us to maximize the optimization quality while the computational cost is kept moderate.
Shinji KITA Seiichi OZAWA Satoshi MAEKAWA Shigeo ABE
In this paper, we present a new method to enhance classification performance of a multiple classifier system by combining a boosting technique called AdaBoost.M2 and Kernel Discriminant Analysis (KDA). To reduce the dependency between classifier outputs and to speed up the learning, each classifier is trained in a different feature space, which is obtained by applying KDA to a small set of hard-to-classify training samples. The training of the system is conducted based on AdaBoost.M2, and the classifiers are implemented by Radial Basis Function networks. To perform KDA at every boosting round in a realistic time scale, a new kernel selection method based on the class separability measure is proposed. Furthermore, a new criterion of the training convergence is also proposed to acquire good classification performance with fewer boosting rounds. To evaluate the proposed method, several experiments are carried out using standard evaluation datasets. The experimental results demonstrate that the proposed method can select an optimal kernel parameter more efficiently than the conventional cross-validation method, and that the training of boosting classifiers is terminated with a fairly small number of rounds to attain good classification accuracy. For multi-class classification problems, the proposed method outperforms both Boosting Linear Discriminant Analysis (BLDA) and Radial-Basis Function Network (RBFN) with regard to the classification accuracy. On the other hand, the performance evaluation for 2-class problems shows that the advantage of the proposed BKDA against BLDA and RBFN depends on the datasets.
Yoshihisa OKADA Tomotaka WADA Masato HORIE Fumio NAKASE Hiromi OKADA
Inter-Vehicle Communication (IVC) is one of the most important technologies to realize advanced Intelligent Transport Systems (ITS). We extensively apply the IVC technology to the communications between pedestrians and vehicles. We call this kind of communications VPEC (Vehicle-PEdestrian Communications). The objective of this paper is to present an effective control scheme for VPEC and to evaluate the performance of proposed scheme by experiments. We deal with direct communications between pedestrians and vehicles. Due to the battery shortage of pedestrians' terminals (p-node), we have presented a reflect-transmission scheme. In this paper, we propose a new access protocol for reflect-transmission scheme, and show its validity by various experiments with several vehicles.
This paper proposes a new binary motion estimation algorithm that improves the motion vector accuracy by using a hybrid distortion measure. Unlike conventional binary motion estimation algorithms, the proposed algorithm considers the sum of absolute difference (SAD) as well as the sum of bit-wise difference (SBD) as a block-matching criterion. In order to reduce the computational complexity and remove additional memory accesses, a new scheme is used for SAD calculation. This scheme uses 8-bit data of the lowest layer already moved into the local buffer to calculate the SAD of other higher binary layer. Experimental results show that the proposed algorithm finds more accurate motion vectors and removes the blockishness of the reconstructed video effectively. We applied this algorithm to existing video encoder and obtained noticeable visual quality enhancement.
Wei-min WANG Du-yan BI Xing-min DU Lin-hua MA
A novel high-speed and area-efficient Reed-Solomon decoder is proposed, which employs pipelining architecture of minimized modified Euclid (ME) algorithm. The logic synthesis and simulation results of its VLSI implementation show that it not only can operate at a higher clock frequency, but also consumes fewer hardware resources.
Shinichiro OMACHI Masako OMACHI Hirotomo ASO
In statistical pattern recognition, it is important to estimate the distribution of patterns precisely to achieve high recognition accuracy. In general, precise estimation of the parameters of the distribution requires a great number of sample patterns, especially when the feature vector obtained from the pattern is high-dimensional. For some pattern recognition problems, such as face recognition or character recognition, very high-dimensional feature vectors are necessary and there are always not enough sample patterns for estimating the parameters. In this paper, we focus on estimating the distribution of high-dimensional feature vectors with small number of sample patterns. First, we define a function, called simplified quadratic discriminant function (SQDF). SQDF can be estimated with small number of sample patterns and approximates the quadratic discriminant function (QDF). SQDF has fewer parameters and requires less computational time than QDF. The effectiveness of SQDF is confirmed by three types of experiments. Next, as an application of SQDF, we propose an algorithm for estimating the parameters of the normal mixture. The proposed algorithm is applied to face recognition and character recognition problems which require high-dimensional feature vectors.
It is predicted that there will be a high demand for video applications in future wireless networks including wireless ad hoc networks. However, supporting video applications over mobile ad hoc networks is more complicated than with other networks due to the lack of support from a preinstalled infrastructure. In this paper, we tackle this problem by adopting the concept of multipoint-to-point video transmission used in wire-line networks. A novel framework designed with features to accommodate the characteristics of ad hoc networks is presented. There are two key features in our proposal. First, Multiple Description Coding (MDC) scheme is used for video coding to reduce the redundancy by avoiding the transmission of duplicate video frames. Second, the routing protocol is expanded to include finding disjoint routes from video sources to the receiver so that a single link breakage or a single intermediate node failure affects transmission from only the minimum number of nodes. Furthermore, the use of disjoint routes also enables the workload to be distributed more evenly within the network. A simulation study was carried out using NS-2 to demonstrate the performance of the proposed mechanism. We show that the proposed mechanism outperforms conventional point-to-point transmission, especially under conditions of high mobility.
Masayoshi NAITO Yohko MICHIOKA Kuniaki OZAWA Yoshitoshi ITO Masashi KIGUCHI Tsuneo KANAZAWA
A communication means is presented for patients with amyotrophic lateral sclerosis in totally locked-in state who are completely unable to move any part of the body and have no usual communication means. The method utilizes changes in cerebral blood volume accompanied with changes in brain activity. When a patient is asked a question and the answer to it is 'yes', the patient makes his or her brain active. The change in blood volume at the frontal lobe is detected with near-infrared light. The instantaneous amplitude and phase of the change are calculated, and the maximum amplitude and phase change are obtained. The answer 'yes' or 'no' of the patient is detected using a discriminant analysis with these two quantities as variables. The rate of correct detection is 80% on average.
Estimating the generalization error is one of the key ingredients of supervised learning since a good generalization error estimator can be used for model selection. An unbiased generalization error estimator called the subspace information criterion (SIC) is shown to be useful for model selection, but its range of application is limited to linear learning methods. In this paper, we extend SIC to be applicable to non-linear learning.
Shiunn-Jang CHERN Chun-Hung SUN
The performance of the blind multiuser detector for a DS-CDMA system with linearly constrained constant modulus (LCCM) criterion is known to highly depend on the exact knowledge of the desired user amplitude; it is usually not available at receiver end. In this paper, we propose a novel LC adaptive CM RLS (LC-ACM-RLS) algorithm to adaptively implement the optimal solution of the LCCM receiver, and to track the desired user's amplitude, simultaneously. From computer simulations, we verify the superiority of the new proposed algorithm over the conventional LCCM-RLS algorithm for multiple access interference (MAI) suppression. Also, for time-varying channel during the adaptation processes, if the amplitude of desired user is not available and varies with time, such as hand-off and Rayleigh fading environments, we show that the proposed LC-ACM-RLS algorithm has better tracking capability compared with the conventional approaches.