Unstructured overlay networks are widely adopted in large-scale and heterogeneous peer-to-peer (P2P) systems for their scalability and flexibility. A distinct feature of such systems is that they randomly route messages e.g., by flooding or random walk. In such systems, the number of messages and tasks carrying by those messages each peer receives is greatly affected by the number of the peer's incoming links. The objective of this paper is to build controllable degree-weighted networks in which the expected number of incoming links of each peer is proportional to its weight which is a local parameter. In such a network, a peer can control the number of those randomly disseminated messages and tasks it receives by adjust it weight. In addition, in order to bound the construction overhead for highly biased networks, we restrict all peers to have the same number of outgoing links. The objective network is constructed by local topology transformations that peers periodically exchange outgoing links with each other. A framework, which includes 81 different protocols by combination of exchange rules, is presented and evaluated by simulation. The simulation result shows that two of them can generate the networks having similar properties with the objective network. This work first achieves the weight-proportional degree control under the out-regular network model.

View full abstract

The regular 2-D mesh topology has been utilized for most of Network-on-Chips (NoCs) on FPGAs. Spatially biased traffic generated in some applications makes a customization method for removing links more efficient, since some links become low utilization. In this paper, a link removal strategy that customizes the router in NoC is proposed for reconfigurable systems in order to minimize the required hardware amount. Based on the pre-analyzed traffic information, links on which the communication amount is small are removed to reduce the hardware cost while maintaining adequate performance. Two policies are proposed to avoid deadlocks and they outperform up*/down* routing, which is a representative deadlock-free routing on irregular topology. In the case of the image recognition application susan, the proposed method can save 30% of the hardware amount without performance degradation.

View full abstract

In the arsenal of resources for computer memory system performance improvement, predictors have gained an increasing role in the past years. They can suppress the latencies when accessing cache or main memory. In paper[1] it is shown how temporal parameters of cache memory access, defined as live time, dead time and access interval could be used for prediction of data prefetching. This paper examines the feasibility of applying an analog technique on controlling of opening/closing DRAM memory rows, with various improvements. The results described herein confirm the feasibility, and allow us to propose a DRAM controller with predictors that not only close the opened DRAM row, but also predict the next row to be opened.

View full abstract

Finding an appropriate data transformation between two schemas has been an important problem. In this paper, assuming that an update script between original and updated DTDs is available, we consider inferring a transformation algorithm from the original DTD and the update script such that the algorithm transforms each document valid against the original DTD into a document valid against the updated DTD. We first show a transformation algorithm inferred from a DTD and an update script. We next show a sufficient condition under which the transformation algorithm inferred from a DTD d and an update script is unambiguous, i.e., for any document t valid against d, elements to be deleted/inserted can unambiguously be determined. Finally, we show a polynomial-time algorithm for testing the sufficient condition.

View full abstract

In this paper, an information distribution model based on human's behavior is proposed. We also propose dynamic parameters to make the model more practical for real life social network. Subsequently, the simulations are conducted based on the small-world network and its characteristics, and the parameters in the model are analyzed to increase efficiently the power of information distribution. Our study and simulation results show that the proposed model can be used to analyze and predict the effectiveness of information distribution. Moreover, the study also shows how to use the model parameters to increase power of the distribution.

View full abstract

In static wireless ad hoc networks such as wireless mesh networks and wireless sensor networks, multipath routing techniques are very useful for improving end-to-end delay, throughput, and load balancing, as compared to single-path routing techniques. When determining multiple paths, however, multipath routing protocols should address the well-known route coupling problem that results from a geographic proximity of adjacent routes and that hampers performance gain. Although a lot of multipath routing protocols have been proposed, most of them focused on obtaining node or link-disjoint multipaths. In order to address the route coupling problem, some multipath routing protocols utilizing zone-disjointness property were proposed. However, they suffer from an overhead of control traffic or require additional equipment such as directional antenna. This paper therefore proposes a novel multipath routing protocol, based on geographical information with low overhead, called 3-directional zone-disjoint multipath routing protocol (3DMRP). 3DMRP searches up to three zone-disjoint paths by using two techniques: 1) greedy forwarding, and 2) RREP-overhearing. One primary and two secondary paths are obtained via greedy forwarding in order to reduce control overhead, and these secondary paths are found by avoiding the RREP overhearing zone created during the primary path acquisition. In particular, two versions of 3DMRP are introduced in order to avoid the RREQ-overhearing zone. Through ns-2 simulations, 3DMRP is evaluated to verify that it achieves performance improvements in terms of throughput and control overhead.

View full abstract

This paper focuses on the collaborative learning of mathematics in which learners effectively acquire knowledge of common exercises through discussion with other learners. During collaborative learning, learners sometimes cannot solve exercises successfully, because they cannot derive answers by themselves or they hesitate to propose answers through discussion. To cope with such situations, this paper proposes two support functions using diagrams to encourage active discussion, since diagrams are often used to graphically illustrate mathematical concepts. One function indicates the differences between learner diagrams and the group diagram in order to encourage participation in discussions. To compare the characteristics of diagrams drawn by different learners, internal representation of the diagram, which consists of types of figures and remarkable relations to other figures, is introduced. The other function provides hints in the group diagram so that all learners can consider their answers collaboratively through discussions. Since preparing hints for all exercises is difficult, rules for drawing supplementary figures, which are general methods for drawing supplementary figures that correspond to individual answering methods/formulas, are also developed. By applying available rules to current group diagram, appropriate supplementary figures that can solve current learning situations may be generated. The experimental results showed that the generated hints successfully increased the number of utterances in the groups. Moreover, learners were also able to derive answers by themselves and tended to propose more opinions in discussions when the uniqueness of their diagrams was indicated.

View full abstract

We propose an appearance manifold with view-dependent covariance matrix for face recognition from video sequences in two learning frameworks: the supervised-learning and the incremental unsupervised-learning. The advantages of this method are, first, the appearance manifold with view-dependent covariance matrix model is robust to pose changes and is also noise invariant, since the embedded covariance matrices are calculated based on their poses in order to learn the samples' distributions along the manifold. Moreover, the proposed incremental unsupervised-learning framework is more realistic for real-world face recognition applications. It is obvious that it is difficult to collect large amounts of face sequences under complete poses (from left sideview to right sideview) for training. Here, an incremental unsupervised-learning framework allows us to train the system with the available initial sequences, and later update the system's knowledge incrementally every time an unlabelled sequence is input. In addition, we also integrate the appearance manifold with view-dependent covariance matrix model with a pose estimation system for improving the classification accuracy and easily detecting sequences with overlapped poses for merging process in the incremental unsupervised-learning framework. The experimental results showed that, in both frameworks, the proposed appearance manifold with view-dependent covariance matrix method could recognize faces from video sequences accurately.

View full abstract

The goal of personal-style handwriting synthesis is to produce texts in the same style as an individual writer by analyzing the writer's samples of handwriting. The difficulty of handwriting synthesis is that the output should have the characteristics of the person's handwriting as well as looking natural, based on a limited number of available examples. We develop a synthesis algorithm which produces handwriting that exhibits naturalness based on the probabilistic character model.

View full abstract

In this paper we describe a new framework of feature combination in the cepstral domain for multi-input robust speech recognition. The general framework of working in the cepstral domain has various advantages over working in the time or hypothesis domain. It is stable, easy to maintain, and less expensive because it does not require precise calibration. It is also easy to configure in a complex speech recognition system. However, it is not straightforward to improve the recognition performance by increasing the number of inputs, and we introduce the concept of variance re-scaling to compensate the negative effect of averaging several input features. Finally, we propose to take another advantage of working in the cepstral domain. The speech can be modeled using hidden Markov models, and the model can be used as prior knowledge. This approach is formulated as a new algorithm, referred to as Hypothesis-Based Feature Combination. The effectiveness of various algorithms are evaluated using two sets of speech databases. We also refer to automatic optimization of some parameters in the proposed algorithms.

View full abstract

This paper describes a distinctive phonetic feature (DPF) extraction method for use in a phoneme recognition system; our method has a low computation cost. This method comprises three stages. The first stage uses two multilayer neural networks (MLNs): MLN_LF-DPF, which maps continuous acoustic features, or local features (LFs), onto discrete DPF features, and MLN_Dyn, which constrains the DPF context at the phoneme boundaries. The second stage incorporates inhibition/enhancement (In/En) functionalities to discriminate whether the DPF dynamic patterns of trajectories are convex or concave, where convex patterns are enhanced and concave patterns are inhibited. The third stage decorrelates the DPF vectors using the Gram-Schmidt orthogonalization procedure before feeding them into a hidden Markov model (HMM)-based classifier. In an experiment on Japanese Newspaper Article Sentences (JNAS) utterances, the proposed feature extractor, which incorporates two MLNs and an In/En network, was found to provide a higher phoneme correct rate with fewer mixture components in the HMMs.

View full abstract

Context-based Adaptive Binary Arithmetic Coding(CABAC) is one of the algorithmic improvements that the H.264/AVC standard provides to enhance the compression ratio of video sequences. Compared with the context-based adaptive variable length coding (CAVLC), CABAC can obtain a better compression ratio at the price of higher computation complexity. In particular, the inherent data dependency and various types of syntax elements in CABAC results in a dramatically increased complexity if two bins obtained from binarized syntax elements are handled at a time. By analyzing the distribution of binarized bins in different video sequences, this work shows how to effectively improve the encoding rate with limited hardware overhead by allowing only a certain type of syntax element to be processed two bins at a time. Together with the proposed context memory management scheme and range renovation method, experimental results reveal that an encoding rate of up to 410M-bin/s can be obtained with a limited increase in hardware requirement. Compared with related works that do not support multi-symbol encoding, our development can achieve nearly twice their throughput rates with less than 25% hardware overhead.

View full abstract

Noise removal in engineering drawing is an important operation performed before other image analysis tasks. Many algorithms have been developed to remove salt-and-pepper noise from document images. Cleaning algorithms should remove noise while keeping the real part of the image unchanged. Some algorithms have disadvantages in cleaning operation that leads to removing of weak features such as short thin lines. Others leave the image with hairy noise attached to image objects. In this article a noise removal procedure called TrackAndMayDel (TAMD) is developed to enhance the noise removal of salt-and-pepper noise in binary images of engineering drawings. The procedure could be integrated with third party algorithms' logic to enhance their ability to remove noise by investigating the structure of pixels that are part of weak features. It can be integrated with other algorithms as a post-processing step to remove noise remaining in the image such as hairy noise attached with graphical elements. An algorithm is proposed by incorporating TAMD in a third party algorithm. Real scanned images from GREC'03 contest are used in the experiment. The images are corrupted by salt-and-pepper noise at 10%, 15%, and 20% levels. An objective performance measure that correlates with human vision as well as MSE and PSNR are used in this experiment. Performance evaluation of the introduced algorithm shows better-quality images compared to other algorithms.

View full abstract

We propose a learning and control model of the arm for a loading task in which an object is loaded onto one hand with the other hand, in the sagittal plane. Postural control during object interactions provides important points to motor control theories in terms of how humans handle dynamics changes and use the information of prediction and sensory feedback. For the learning and control model, we coupled a feedback-error-learning scheme with an Actor-Critic method used as a feedback controller. To overcome sensory delays, a feedforward dynamics model (FDM) was used in the sensory feedback path. We tested the proposed model in simulation using a two-joint arm with six muscles, each with time delays in muscle force generation. By applying the proposed model to the loading task, we showed that motor commands started increasing, before an object was loaded on, to stabilize arm posture. We also found that the FDM contributes to the stabilization by predicting how the hand changes based on contexts of the object and efferent signals. For comparison with other computational models, we present the simulation results of a minimum-variance model.

View full abstract

A new in vitro experimental tool was developed to study the mechanism of the ultrasonic closing clicks' genesis of mechanical heart valves. Since the newly developed tester adopted compressed air flow directly instead of the blood analog fluid to drive the mechanical heart valve, it is not possibe to generate any cavitation. Closing clicks were measured with a small accelerometer at the surface of the valve holder made of silicone rubber. Ultrasonic closing clicks as well as audible closing clicks, similar to those measured clinically, could be observed using this setup. Thus, it was confirmed that the ultrasonic closing clicks can be generated without the existence of cavitation. Simultaneous measurements of the valve motion were made with a high-speed video camera, and the analysis of the video frames and clicks showed that higher frequency signal components of more than 50kHz could be generated only at the instant of the closure, which means the collision of the occluder with the housing. Eighteen miniature accelerometers with an area of one square millimeter were developed and stuck on the housing to monitor the distribution of the housing vibrations in detail, and it was found that the vibrations correspond to the ultrasonic closing clicks propagated from the valve stop: the collision point of the occluder with the housing. This fact indicated that the generation of ultrasonic closing clicks are limited to the small area of the collision. From those results, it was concluded that the major origin of the ultrasonic closing clicks' genesis should be the collision of the occluder with the housing.

View full abstract

We propose a polynomial-time algorithm for the scheduling of real-time parallel tasks on multicore processors. The proposed algorithm always finds a feasible schedule using the minimum number of processing cores, where tasks have properties of linear speedup, flexible preemption, arbitrary deadlines and arrivals, and parallelism bound. The time complexity of the proposed algorithm is O(M³· log N) for M tasks and N processors in the worst case.

View full abstract

We propose a Solid-State Disk (SSD) with a Double Data Rate (DDR) DRAM interface for high-performance PCs. Traditional SSDs simply inherit the interface protocol of Hard Disk Drives (HDD) such as Parallel Advanced Technology Attachment (PATA) or Serial-ATA (SATA) for maintaining the compatibility. However, SSD itself provides much higher performance than HDD, hence the interface also needs to be enhanced. Unlike the traditional SSDs, the proposed SSD with DDR DRAM interface is placed in the North Bridge which provides two or more DDR DRAM interface ports in high-performance PCs. The novelty of our work is on DQS signaling scheme which allows arbitrary Column Address Strobe (CAS) latency unlike typical DDR DRAM interface scheme. The experimental results show that the proposed SSD maximally outperforms the traditional SSD by 8.7 times in read mode, by 1.5 times in write mode. Also, for synthetic workloads, the proposed scheme shows performance improvement over the conventional architecture by a factor of 1.6 times.

View full abstract

In this paper, we propose an effective Web cache admission control algorithm. By selectively admitting objects into the cache, the proposed algorithm can significantly reduce the amount of disk I/O on a Web cache while maintaining a high hit ratio. The proposed algorithm adaptively adjusts its own admission control parameter, requiring no user-supplied parameters. Through extensive experiments, we show the effectiveness of the proposed algorithm.

View full abstract

Previous approaches to question retrieval in community-based question answering rely on statistical translation techniques to match users' questions (queries) against collections of previously asked questions. This paper presents a simple but effective method for computing word relatedness to improve question retrieval based on word co-occurrence information directly extracted from question and answer archives. Experimental results show that the proposed approach significantly outperforms translation-based approaches.

View full abstract

Privacy preserving is indispensable in data mining. In this paper, we present a novel clustering method for distributed multi-party data sets using orthogonal transformation and data randomization techniques. Our method can not only protect privacy in face of collusion, but also achieve a higher level of accuracy compared to the existing methods.

View full abstract

With proliferation of smart handsets capable of mobile Internet, the severity of malware attacks targeting such handsets is rapidly increasing, thereby requiring effective countermeasure for them. However, existing signature-based solutions are not suitable for resource-poor handsets due to the excessive run-time overhead of matching against ever-increasing malware pattern database as well as the limitation of detecting well-known malware only. To overcome these drawbacks, we present a bio-inspired approach to discriminate malware (non-self) from normal programs (self) by replicating the processes of biological immune system. Our proposed approach achieves superior performance in terms of detecting 83.7% of new malware or their variants and scalable storage requirement that grows very slowly with inclusion of new malware, making it attractive for use with mobile handsets.

View full abstract

In this letter, we propose a new approach to SVM based speaker recognition, which utilizes a kind of novel phonotactic information as the feature for SVM modeling. Gaussian mixture models (GMMs) have been proven extremely successful for text-independent speaker recognition. The GMM universal background model (UBM) is a speaker-independent model, each component of which can be considered as modeling some underlying phonetic sound classes. We assume that the utterances from different speakers should get different average posterior probabilities on the same Gaussian component of the UBM, and the supervector composed of the average posterior probabilities on all components of the UBM for each utterance should be discriminative. We use these supervectors as the features for SVM based speaker recognition. Experiment results on a NIST SRE 2006 task show that the proposed approach demonstrates comparable performance with the commonly used systems. Fusion results are also presented.

View full abstract

Various scanning algorithms have been proposed to enhance the performance of intra prediction of H.264 codec. In this paper, an adaptive scanning scheme is proposed to achieve the entropy coding gain in intra coding, where scanning patterns are updated based on the probabilistic distribution of quantized coefficients of previous macroblocks so that the consecutive zeros are located at the rear part of the scanned data stream. Simulation results show that the average bit-rate reduction is about 2.15% in common test conditions.

View full abstract

Salient Region Extraction provides an alternative methodology to image description in many applications such as adaptive content delivery and image retrieval. In this paper, we propose a robust approach to extracting the salient region based on bottom-up visual attention. The main contributions are twofold: 1) Instead of the feature parallel integration, the proposed saliencies are derived by serial processing between texture and color features. Hence, the proposed approach intrinsically provides an alternative methodology to model attention with low implementation complexity. 2) A constructive approach is proposed for rendering an image by a non-linear intensity mapping, which can efficiently eliminate high contrast noise regions in the image. And then the salient map can be robustly generated for a variety of nature images. Experiments show that the proposed algorithm is effective and can characterize the human perception well.

View full abstract

A new hierarchical isosurface reconstruction scheme from a set of tomographic cross sectional images is presented. From the input data, we construct a hierarchy of volume, called the volume pyramid, based on a 3D dilation filter. After extracting the base mesh from the volume at the coarsest level by the cell-boundary method, we iteratively fit the mesh to the isopoints representing the actual isosurface of the volume. The SWIS (Shrink-wrapped isosurface) algorithm is adopted in this process, and a mesh subdivision scheme is utilized to reconstruct fine detail of the isosurface. According to experiments, our method is proved to produce a hierarchical isosurface which can be utilized by various multiresolution algorithms such as interactive visualization and progressive transmission.

View full abstract