Displaying a 3D geometric model of a user in real time is an advantage for a telecommunication system because depth information is useful for nonverbal communication such as finger-pointing and gesturing that contain 3D information. However, the range image acquired by a rangefinder suffers from errors due to image noises and distortions in depth measurement. On the other hand, a 2D image is free from such errors. In this paper, we propose a new method for a shared space communication system that combines the advantages of both 2D and 3D representations. A user is represented as a 3D geometric model in order to exchange nonverbal communication cues. A background is displayed as a 2D image to give the user adequate information about the environment of the remote site. Additionally, a high-resolution texture taken by a video camera is projected onto the 3D geometric model of the user. This is done because the low resolution of the image acquired by the rangefinder makes it difficult to exchange facial expressions. Furthermore, to fill in the data occluded by the user, old pixel values are used for the user area in the 2D background image. We have constructed a prototype of a high presence shared space communication system based on our method. Through a number of experiments, we have found that our method is more effective for telecommunication than a method with only a 2D or 3D representation.

Currently, multi-party video conference does not provide equivalent quality in comparison to face-to-face conference. One assumed reason is that participants cannot be aware of "who is focusing on whom". We introduce virtual space to a multi-party conference system, allocating avatars in a space. We also introduce intuitive input interface using motion processor in order to construct a multi-party conference system, which the user can use without being aware of it. A new displaying method is essential for this system, and we introduce a way by which a user can obtain the feedback of which user he/she is focusing on. We introduce e-MulCS as the system that fulfils these proposals. By comparing this system with the video conference system, the results show that our system supports the intuitive multi-party communication better.

Due to the cost of multicast state management, multicast address allocation, inter-domain multicast routing of traditional IP multicast scheme, ASM leads to a search for other multicast schemes. This paper presents a new solution to the problems mentioned above based on IPv6. The proposed scheme provides an enhanced scheme supporting the strengths of SSM in basic Xcast. This is achieved by adding MLDv2 operations at recipient's side and a new control plane into existing Xcast. The proposed scheme does not only provide the transparency of traditional multicast schemes to sources and recipients, but it also enhances the routing efficiency in networks. Intermediate routers do not have to maintain multicast state, so that it results in a more efficient and scalable mechanism to deliver native multicast datagrams. Also, the seamless integration in Mobile IPv6 can support multicast efficiently for mobile nodes in IPv6 networks by avoiding tunnel avalanches and tunnel convergence. We've attempted to prove this alternative architecture by both simulation and implementation, respectively. Our approach cannot fundamentally perform for many large groups distributed widely as effectively as traditional multicast schemes. However, we believe that the resulting scheme is simple, efficient, robust, transparent, and to the extent possible, scalable in case that recipients are clustered in subnets.

Mobile IP provides an efficient and scalable mechanism for host mobility within the Internet. Using Mobile IP, mobile nodes may change their point of attachment to the Internet without changing their IP address. In contrast to the advantages of Mobile IP, updating the location of a mobile node incurs high signaling costs if the mobile node moves frequently. Thus, IP paging schemes for Mobile IP have been proposed to avoid unnecessary registration signaling overhead when a mobile node is idle. However, they require the additional paging costs and delays associated with message delivery when a correspondent node sends packets to the idle mobile node. These delays greatly influence the quality of service (QoS) for multimedia services. Moreover, existing IP paging schemes do not consider a hierarchical mobility management scheme, which can reduce signaling costs by the significant geographic locality in user mobility pattern. Thus, we propose a novel IP paging protocol which can be used in hierarchical Mobile IPv6 networks. In addition, our proposal can reduce signaling costs for paging and delay using the concept of the anchor-cell. The cost analysis presented in this paper shows that our proposal has superior performance when the session-to-mobility ratio value of the mobile node is low.

AF service class in Diffserv by realizes minimum bandwidth guarantee with the use of differentiated drop precedence property marked on each packet. In the context of a multiple domains environment, however, QoS of individual flow is not always preserved due to the re-marking behavior forced at the domain boundaries. Focusing on this point, this paper proposes new packet re-marking schemes that can improve the per-flow QoS of AF service traversing multiple domains. The basic concept of the schemes distinguishes packets re-marked to out-of-profile at the domain boundaries from those already marked as out-of-profile at the time of entering the network, and allows the re-marked packet to recover back to in-of-profile, thus regaining its rightful QoS within the networks. The performance of the proposed schemes is evaluated through simulation. The results on UDP flows show the effectiveness of the proposed schemes for reducing packet losses on the flow through multiple domains and preserving fairness between flows. Simulations on TCP flows show that the proposed schemes improve the throughput of the flows through multiple domains. The proposed scheme is especially effective on the transfer time of short TCP flows such as Web traffic, whose throughput is affected more seriously by a single packet loss due to its flow control mechanisms.

In IP-based mobile networks, a few of mobility agents (e.g., home agent, foreign agent, etc.) are used for mobility management. Recently, Hierarchical Mobile IPv6 (HMIPv6) was proposed to reduce signaling overhead and handoff latency occurred in Mobile IPv6. In HMIPv6, a new mobility agent, called mobility anchor point (MAP), is deployed in order to handle binding update procedures locally. However, the MAP can be a single point of performance bottleneck when there are a lot of mobile node (MNs) performing frequent local movements. This is because the MAP takes binding update procedures as well as data packet tunneling. Therefore, it is required to control the number of MNs serviced by a single MAP. In this paper, we propose a load control scheme at the MAP utilizing an admission control algorithm. We name the proposed load control scheme proactive load control scheme to distinct from the existing load control schemes in cellular networks. In terms of admission control, we use the cutoff priority scheme. We develop Markov chain models for the proactive load control scheme and evaluate the ongoing MN dropping and the new MN blocking probabilities. As a result, the proactive load control scheme can reduce the ongoing MN dropping probability while keeping the new MN blocking probability to a reasonable level.

Many P2P lookup services based on distributed hash tables (DHT) have appeared recently. These schemes are built upon overlay networks and ignore distance to the target resources. As a result, P2P lookups often suffer from unnecessarily long routes in the underlay network, which we call overlay dilation. This paper proposes a new scheme for resource routing, called hybrid hierarchical overlay routing, dubbed Hyho. We introduce distance-weighted Bloom filters (dwBFs) as a concise representation of routing information for scattered resources in overlay networks. To further reduce the size of Bloom filters, so that they are linear in the number of distinct resources, Hyho splits overlay networks in accordance with DHT, where each subnetwork has a smaller set of resources and spans the entire network thinly. As a result, Hyho constructs a hierarchical overlay network and routes requests accordingly. Simulation results show that Hyho can reduce overlay dilation to one half that yielded by the Chord lookup service.

In this paper, we propose an MPEG-2 TS encryption and decryption scheme which can support MPEG-2 TS streaming, and a key management scheme which can provide secure delivery of the key used to encrypt MPEG-2 TS from the package server to DRM clients in a secure manner. The key management scheme protects the key not only from purchasers but also among the other principals who manage the distribution servers and license servers. The proposed scheme can protect from attacks over a network since a pre-encrypted MPEG-2 TS is sent by a streaming server and only DRM clients can decrypt the streamed MPEG-2 TS. An MPEG-2 TS streaming server can send the encrypted stream without modifying it and degrading its performance because MPEG-2 TS is encrypted but compliant to the MPEG-2 TS format. The processing time to decrypt a streamed MPEG-2 TS in client systems is minimized to guarantee the quality of streaming.

Source-Specific Multicast (SSM) is recognized as the most feasible multicast communication model for a wide use throughout the Internet because it eliminates many complexities associated with traditional many-to-many multicast communication model. Regarding the SSM communication, however, implementations of the Internet Group Multicast Protocol version 3 (IGMPv3) for IPv4 and the Multicast Listener Discovery version 2 (MLDv2) for IPv6 are indispensable on every end-host. Accurate implementations of these protocols are necessary studies for the smooth deployment. In this document, we illustrate IGMPv3 and MLDv2 host-side implementations applied to the 4.4BSD kernel. While presenting the concept and the design of these implementations, we verify the behavior and evaluate the response times for (^*,G) and (S,G) joins requested on top of both implementations.

Availability of network access "anytime and anywhere" will impose new requirements to presence services - server load sharing and privacy protection. In such cases, presence services would have to deal with sensor device information with maximum consideration of user's privacy. In this paper, we propose FieldCast: peer-to-peer system architecture for presence information exchange in ubiquitous computing environment. According to our proposal, presence information is exchanged directly among user's own computing resources. We illustrate our result of evaluation that proves the feasibility of our proposal.

This innovative traffic-monitoring-system makes it possible to observe data-communication traffic on an oscilloscope-style display. It provides an efficient way of evaluating streaming-data quality. The monitoring system has a high time-resolution traffic value sampling function and a real-time data representation/recording mechanism that operate in synchrony. The user can directly evaluate the traffic shape with the monitoring system. In this paper, after describing the concept of the traffic monitoring system, we will describe a prototype built with programmable network equipment called A-BOX. We will then review a performance evaluation and other experimental results to prove that our monitoring system is suitable for video streaming.

Performance measurements are indispensable for managing the Internet. Among the performance measurement techniques known, passive measurement is attractive because of its accuracy; user traffic is observed without inserting additional test traffic. However, the technique is handicapped by its large storage and bandwidth costs. This paper proposes a passive packet loss measurement technique that effectively avoids the difficulty of the conventional passive measurement approaches. Its key advance is utilizing the packet feature computed by a hash function. Since the feature can identify a packet with a short length of data, it becomes possible to greatly decrease the storage and bandwidth costs of passive measurements. The paper details the measurement procedure and assesses the design parameters used in the method. In addition, the validity of the proposed method is confirmed through experiments. The experiments also show the advantage of the method over the conventional active measurement.

In recent years, interruption of services large-scale business sites and Root Name Servers caused by Denial-of-Service (DoS) attacks or Distributed DoS (DDoS) attacks has become an issue. Techniques for specifying attackers are, thus important. On the other hand, since information on attackers' source IP addresses are generally spoofed, tracing techniques are required for DoS attacks. In this paper, we predict network traffic volume at observation points on the network, and detect DoS attacks by carefully examining the difference between predicted traffic volume and actual traffic volume. Moreover, we assume that the duration time of an attack is the same at every observation point the attack traffic passes, and propose a tracing method that uses attack duration time as a parameter. We show that our proposed method is effective in tracing DDoS attacks.

Analysing and modeling of traffic play a vital role in designing and controlling of networks effectively. To construct a practical traffic model that can be used for various networks, it is necessary to characterize aggregated traffic and user traffic. This paper investigates these characteristics and their relationship. Our analyses are based on a huge number of packet traces from five different networks on the Internet. We found that: (1) marginal distributions of aggregated traffic fluctuations follow positively skewed (non-Gaussian) distributions, which leads to the existence of "spikes", where spikes correspond to an extremely large value of momentary throughput, (2) the amount of user traffic in a unit of time has a wide range of variability, and (3) flows within spikes are more likely to be "elephant flows", where an elephant flow is an IP flow with a high volume of traffic. These findings are useful in constructing a practical and realistic Internet traffic model.

It is important to predict serious deterioration of telecommunication quality. This paper investigates predicting such serious events by analyzing only a "short" period (i.e., a "small" amount) of teletraffic data. To achieve this end, this paper presents a method for analyzing the tail distributions of teletraffic state variables, because tail distributions are suitable for representing serious events. This method is based on Extreme Value Theory (EVT), which provides a firm theoretical foundation for the analysis. To be more precise, in this paper, we use throughput data measured on an actual network during daily busy hours for 15 minutes, and use its first 10 seconds (known data) to analyze the tail distribution. Then, we evaluate how well the obtained tail distribution can predict the tail distribution of the remaining 890 seconds (unknown data). The results indicate that the obtained tail distribution based on EVT by analyzing the small amount of known data can predict the tail distribution of unknown data much better than methods based on empirical or log-normal distributions. Furthermore, we apply the obtained tail distribution to predict the peak throughput in unknown data. The results of this paper enable us to predict serious deterioration events with lower measurement cost.

There is an emerging requirement for real-time flow-based traffic monitoring, which is vital to detecting and/or tracing DoS attacks as well as troubleshooting and traffic engineering in the ISP networks. We propose the architecture for a scalable real-time flow measurement tool in order to allow operators to flexibly define "the targeted flows" on-demand, to obtain various statistics on those flows, and to visualize them in a real-time manner. A traffic distribution device and multiple traffic capture devices processing packets in parallel are included in the architecture, in which the former device copies traffic and distributes it to the latter devices. We evaluate the performance of a proto-type implementation on PC-UNIX in testbed experiments to demonstrate the scalability of our architecture. The evaluation shows that the performance increases in proportion to the number of the capture devices and the maximum performance reaches 80 K pps with six capture devices. Finally we also show applications of our tool, which indicate the advantage of flexible fine-grained flow measurements.

The volume of mass unsolicited electronic mail, often known as spam, has recently increased enormously and has become a serious threat not only to the Internet but also to society. This paper proposes a new spam detection method which uses document space density information. Although the proposed method requires extensive e-mail traffic to acquire the necessary information, it can achieve perfect detection (i.e., both recall and precision is 100%) under practical conditions. A direct-mapped cache method contributes to the handling of over 13,000 e-mail messages per second. Experimental results, which were conducted using over 50 million actual e-mail messages, are also reported in this paper.

As audio and video applications have proliferated on the Internet, transcoding proxy caching has been considered as an important technique for improving network performance, especially for mobile networks. Due to several new emerging factors in the transcoding proxy, existing methods for proxy placement for web caching cannot be simply applied to solve the problem of proxy placement for transcoding proxy caching. This paper addresses the problem of proxy placement for coordinated en-route transcoding proxy caching for tree networks. We propose a model for this problem by including the new emerging factors in the transcoding proxy and present optimal solutions for this problem with/without constraints on the number of transcoding proxies using dynamic programming. Finally, we implement our algorithm and evaluate our model on various performance metrics through extensive simulation experiments. The implementation results show that our model outperforms the existing model for transcoding proxy placement for linear topology, as well as the random proxy placement model. The average improvements of our model over the other models are about 7.2 percent and 21.4 percent in terms of all the performance metrics considered.

Replication to mask the effects of failures is a basic technique for improving reliability of a file system. Consistency control protocols are implemented to ensure the consistency among these replicas. The native token-based mechanism which merely sequences the distributed requests suffered from the poor system utilization due to the lack of dependence checking between writes and management of out-of-ordered requests. Hence, in this paper, by utilizing the idempotent property of NFS and executing ahead most of independent WRITE requests, the new consistency control scheme is proposed to improve the performance of operations and failure recovery. Finally, a numeric case shows the efficiency of the new scheme.

The massively multiplayer online game (MMOG) industry is suffering from huge outgoing traffic from centralized servers. To accommodate this traffic, game companies claim large bandwidth to Internet Data Centers (IDCs), and several months' payment for that bandwidth is likely to even exceed the cost for MMOG servers. In this paper, we propose a MMOG server architecture to reduce outgoing bandwidth consumption from MMOG servers. The proposed architecture distributes some functions of servers to selected clients, and those clients are in charge of event notification to other clients in order to reduce the outgoing traffic from servers. The clients with server functions communicate with each other in peer-to-peer manner. We analyze traffic reduction as a function of cell-daemonable ratio of clients, and the results show that up to 80% of outgoing traffic from servers can be reduced using the proposed architecture when 10% of clients are cell-daemonable.

As Internet access had spread, it has become an important potential means of lifeline communication. The Internet can fulfill a role in the everyday life of citizens by offering lifeline communication services such as the police, ambulance, gas, electricity, and water services when an emergency occurs. In these lifeline communications, a caller needs to be able to communicate with the nearby lifeline service in the same manner as on a PSTN (Public Switched Telephone Networks) without being consciously aware of the path the caller's message is taking. Moreover, the lifeline service agency must be able to acquire the caller's location and identity. However, it is very difficult to transfer these essential functions from the PSTN to the Internet without making significant changes, because of large differences between the PSTN and the Internet. In this paper, we discuss how to obtain these functions for lifeline communications in the Internet. We further propose a model and implement a lifeline communication system on the Internet.

CDN (Content Delivery Networks) improves end-user performance by replicating web contents on a group of geographically distributed servers. However, repeatedly keeping the entire replica of the original objects into many content servers consumes too much server resource. This problem becomes more serious for the large-sized objects such as streaming media, e.g. high quality video. In this paper, we therefore propose an efficient replication method for layered video streams in CDN, which can reduce user response delays and storage costs simultaneously. Based on an analytical formulation of the cooperative replication of layers and segments of each video stream, we derive a replication algorithm which solves next three problems quantitatively. (1) How many servers should be selected to replicate a given video stream? (2) For a single video stream, how many layers and segments should be stored in a given server? (3) After selecting a group of servers for each video stream, how do we allocate the replication priority (i.e. order) to each server? Simulation results verify that the proposed algorithm efficiently resolves the above problems and provides much better performance than conventional methods.

In order to respond to the active market's needs for software with various new functions, the system testing must be completed within a limited period. Additionally, important faults, which are closely related to essential functions for users or the target system, have to be removed, preferably in system testing. Many techniques have been proposed to date for effective software testing. Among them, selective software testing is one of the most cost effective techniques. However, most of the previous techniques cannot be applied to short-term development and initial development of software with various new functions because much cost is needed for their testing preparation. In this paper, we propose a new method for selective system testing in which priorities assigned to functions play an essential role in the execution of testing. The priorities are determined based on the evaluation results of three metrics for functions: the frequency of use, the complexity of use scenario, and the fault impact to users. Detailed testing instructions are assigned to test items with high priority, and short and ordinal instructions are assigned to those with low priority. The difference in the volume of testing instruction controls the effort of checking test items. As a result of experimental application to actual software testing in a certain company, we have confirmed that the proposed selective system testing can detect both fatal faults related to key functions and critical faults for the system.

The methodologies of product-line engineering emphasize proactive reuse to construct high-quality products more quickly that are less costly. Requirement engineering for software product families differs significantly from requirement engineering for single software products. The requirements for a product line are written for the group of systems as a whole, with requirements for individual systems specified by a delta or an increment to the generic set. Therefore, it is necessary to identify and explicitly denote the regions of commonality and points of variation at the requirement level. In this paper, we suggest a method of producing requirements that will be a core asset in the product line. Briefly, requirements for families of similar systems (i.e. domain) are collected and generalized which are then analyzed and modeled. The domain requirement as a core asset explicitly manages the commonality and variability. Through this method, the reuse of domain requirements can be enhanced. As a result, the cost and time of software development can be reduced and the productivity increased while significantly reducing error in the requirements.

In this paper, we consider a data mining problem for semi-structured data. Modeling semi-structured data as labeled ordered trees, we present an efficient algorithm for discovering frequent substructures from a large collection of semi-structured data. By extending the enumeration technique developed by Bayardo (SIGMOD'98) for discovering long itemsets, our algorithm scales almost linearly in the total size of maximal tree patterns contained in an input collection depending mildly on the size of the longest pattern. We also developed several pruning techniques that significantly speed-up the search. Experiments on Web data show that our algorithm runs efficiently on real-life datasets combined with proposed pruning techniques in the wide range of parameters.

The most important function of a router is to perform IP lookup that determines the output ports of incoming IP packets by their destination addresses. Hence, IP lookup is one of the main issues in implementing high-speed routers. The IP lookup algorithm implemented in IQ2200 Chipset with two-level table architecture can efficiently use memory. However, it wastes processor resource for full re-construction of the forwarding tables whenever every route insertion/deletion is requested. In order to improve the utilization of processor resource, we propose an IP lookup algorithm with three-level table architecture for high-speed routers. We evaluate the performance of the proposed algorithm in terms of the memory size required for storing lookup information and the number of memory access in constructing forwarding tables. Being compared with the IQ2200 scheme, the proposed scheme can reduce the number of memory access up to 99% even though it needs about 16% more memory.

This paper introduces the problem of n mobile agents that repeatedly visit all n nodes of a given network, subject to the constraint that no two agents can simultaneously occupy a node. This paper first presents a self-stabilizing phase-based protocol for a tree network on a synchronous model. The protocol realizes agent traversal with O(Δn) time where n is the number of nodes and Δ is the maximum degree of any vertex in the communication network. The phase-based protocol can also be applied to an asynchronous model and a ring network. This paper also presents a self-stabilizing link-alternator-based protocol with agent traversal time of O(Δn) for a tree network on an asynchronous model. The protocols are proved to be asymptotically optimal with respect to the agent traversal time.

The growth of the Internet has resulted in an increasing need for personalized information systems. The paper describes an autonomous agent, the Web Robot Agent or WebBot, which integrates with the web and acts as a personal recommendation system that cooperates with the user in order to identify interesting pages. The Apriori algorithm extracts the characteristics of the web pages in the form of association words that are semantically related and mines a bag of association words. Using hybrid components from collaborative filtering and content-based filtering, this hybrid recommendation system can overcome the shortcomings associated with traditional recommendation systems. In this paper, we present an improved recommendation system, which uses the user preference mining through hybrid 2-way filtering. The proposed method was tested on a database, and its effectiveness compared with existent methods was proven in on-line experiments.

This article introduces a novel approach to model phonology and morphosyntax in morpheme unit-based speech recognizers. The proposed methods are evaluated on a Hungarian newspaper dictation task that requires modeling over 1 million different word forms. The architecture of the recognition system is based on the weighted finite-state transducer (WFST) paradigm. The vocabulary units used in the system are morpheme-based in order to provide sufficient coverage of the large number of word-forms resulting from affixation and compounding. Besides the basic pronunciation model and the morpheme N-gram language model we evaluate a novel phonology model and the novel stochastic morphosyntactic language model (SMLM). Thanks to the flexible transducer-based architecture of the system, these new components are integrated seamlessly with the basic modules with no need to modify the decoder itself. We compare the phoneme, morpheme, and word error-rates as well as the sizes of the recognition networks in two configurations. In one configuration we use only the N-gram model while in the other we use the combined model. The proposed stochastic morphosyntactic language model decreases the morpheme error rate by between 1.7 and 7.2% relatively when compared to the baseline trigram system. The proposed phonology model reduced the error rate by 8.32%. The morpheme error-rate of the best configuration is 18% and the best word error-rate is 22.3%.

This paper describes an approach to feature extraction in speech recognition systems using kernel principal component analysis (KPCA). This approach represents speech features as the projection of the mel-cepstral coefficients mapped into a feature space via a non-linear mapping onto the principal components. The non-linear mapping is implicitly performed using the kernel-trick, which is a useful way of not mapping the input space into a feature space explicitly, making this mapping computationally feasible. It is shown that the application of dynamic (Δ) and acceleration (ΔΔ) coefficients, before and/or after the KPCA feature extraction procedure, is essential in order to obtain higher classification performance. Better results were obtained by using this approach when compared to the standard technique.

Borrowing the notion of instantaneous frequency that was developed in the context of time-frequency signal analysis, an instantaneous frequency amplitude spectrum (IFAS) is introduced for estimating fundamental frequency of speech signal in both noiseless and adverse environments. We define harmonicity measure as a quantity that indicates degree of periodical regularity in the IFAS and that shows substantial difference between periodic signal and noise-like waveform. The harmonicity measure is applied to estimate the existence of fundamental frequency. We provide experimental examples to demonstrate the general applicability of the harmonicity measure and apply the proposed procedure to Japanese continuous speech signals. The results show that the proposed method outperforms the conventional methods with or without the presence of noise.

This paper describes a method for detecting multiple overlapping objects from a real-time video stream. Layered detection is based on two processes: pixel analysis and region analysis. Pixel analysis determines whether a pixel is stationary or transient by observing its intensity over time. Region analysis detects stationary regions of stationary pixels corresponding to stopped objects. These regions are registered as layers on the background image, and thus new moving objects passing through these layers can be detected. An important aspect of this work derives from the observation that legitimately moving objects in a scene tend to cause much faster intensity transitions than changes due to lighting, meteorological, and diurnal effects. The resulting system robustly detects objects at an outdoor surveillance site. For 8 hours of video evaluation, a detection rate of 92% was measured, which is higher than traditional background subtraction methods.

The problem of signal restoration in the presence of observation space noise has been tackled extensively. However, restoration of degraded signals in the presence of signal space noise leads to considerable complexity because it becomes difficult to distinguish between the original signal and the noise. In this paper, a partial projection filter has been devised for the restoration of signals degraded by both the signal space and the observation space noises. A closed form of the proposed filter has been derived and its performance has been verified experimentally.

The partial projection filter gives optimal signal restoration in the presence of both the signal space and the observation space noises. In this paper, the filter has been characterized from the point of view of its signal restoration and noise suppression capabilities. The filter is shown to suppress the noise component in the restored signal while retaining the signal component, thus maximizing the signal-to-noise ratio. Further, a digital implementation of the filter is presented in matrix form in contrast to its original operator based derivation, for practical applications.

This study proposes an edge-based single-resolution compression scheme for triangular mesh connectivity. The proposed method improves upon EdgeBreaker. Nearly all of these algorithms are either multiple traversals or operate in reverse order. Operating in reverse order should work only off-line in the EdgeBreaker decompression process. Many restrictions on applications will be caused by these factors. To overcome these restrictions, the algorithm developed here can both encode and decode 3D models in a straightforward manner by single traversal in sequential order. Most algorithms require complicated operations when the triangular mesh is split. This study investigates spatial locality to minimize costs in split operations. Meanwhile, some simplification rules are proposed by considering geometric characteristics which ignore the last triangle when a split occurs. The proposed method improves not only the compression ratio but also the execution time.

To implement a fast and reliable question-answering system in Korean, we propose a two-pass answer indexer using co-occurrence information between answer candidates and adjacent content words. The two-pass indexer scans documents twice for obtaining local scores and global scores. Then, the two-pass indexer calculates the degrees of association between answer candidates and co-occurring content words. Using this technique, the proposed QA system shortens the response time and enhances the precision.

This study aims to identify grammatical relations (GRs) in Korean sentences. The key task is to find the GRs in sentences in terms of such GR categories as subject, object, and adverbial. To overcome this problem, we are faced with the structural ambiguity and the grammatical relational ambiguity. We propose a statistical model, which resolves the grammatical relational ambiguity first, and then resolves structural ambiguity by using the probabilities of the GRs given noun phrases and verb phrases in sentences. The proposed model uses the characteristics of the Korean language such as distance, no-crossing and case property. We showed that consideration of such characteristics produces better results than without consideration by experiments. We attempt to enhance our system by estimating the probabilities of the proposed model with the Maximum Entropy (ME) model, and with Support Vector Machines (SVM) classifiers and we confirm that SVM classifiers improved the performance of our proposed model through experiments. Through an experiment with a tree and GR tagged corpus for training the model, we achieved an overall accuracy of 84.8%, 94.1%, and 84.8% in identifying subject, object, and adverbial relations in sentences, respectively.

An important issue in applying machine learning algorithms to Natural Language Processing areas such as Named Entity Recognition tasks is to overcome the lack of tagged corpora. Several bootstrapping methods such as co-training have been proposed as a solution. In this paper, we present a different approach using the Web resources. A Named Entity (NE) tagged corpus is generated from the Web using about 3,000 names as seeds. The generated corpus may have a lower quality than the manually tagged corpus but its size can be increased sufficiently. Several features are developed and the decision list is learned using the generated corpus. Our method is verified by comparing it to both the decision list learned on the manual corpus and the DL-CoTrain method. We also present a two-level classification by cascading highly precise lexical patterns and the decision list to improve the performance.

We describe a new inconsistent case which is susceptible to occur while producing consistent answer set using prioritized default logic. We define new semantics for prioritized default logic in order to solve this problem. There is a sign difference between General and Extended logic programs. Extended logic programs are formulated using classical negation, For this reason, an inconsistent answer set can sometimes be produced. For the most part, default reasoning semantics successfully resolved this problem, but a conflict could still arise in one particular case. The purpose of this paper is to present this eventuality, and revise the semantics of default logic in order to give an answer to this problem.

In this letter, we propose an approach to incorporate a statistical model for the voiced/unvoiced (V/UV) speech decision under background noise environments. Our approach consists of splitting the input noisy speech into two separate bands and applying a statistical model for each band. We compute and compare the likelihood ratio (LR) for each band based on the statistical model and estimated noise statistics for the V/UV decision. According to the simulation test, the proposed V/UV decision shows a better performance compared with the selectable mode vocoder (SMV) V/UV decision algorithm, particularly in clean and white noise environments.

Digital halftoning is used to quantize a grayscale image to a binary image. Error diffusion halftoning generates a high-quality binary image, but also generates some defects such as the warm effect, sharpening, and so forth. To reduce these defects, Kite proposed a modified threshold modulation method that utilizes a multiplicative parameter for controlling sharpening. Nevertheless, some degradation was observed near the edges of objects with a large luminance change. In this paper, we propose a method of controlling the multiplicative parameter in proportion to the magnitude of the local edge slope. The results of computer simulation show a greater reduction of sharpening in the halftone image. In particular, there is a great improvement in the quality of the edges of objects with a large luminance change.

In this paper, we propose a probabilistic feature-based parsing model for head-final languages, which can lead to an improvement of syntactic disambiguation while reducing the parsing cost related to lexical information. For effective syntactic disambiguation, the proposed parsing model utilizes several useful features such as a syntactic label feature, a content feature, a functional feature, and a size feature. Moreover, it is designed to be suitable for representing word order variation of non-head words in head-final languages. Experimental results show that the proposed parsing model performs better than previous lexicalized parsing models, although it has much less dependence on lexical information.

This paper presents a method for improving the performance of syntactic analysis by using accurate temporal expression processing. Temporal expression causes parsing errors due to its syntactic duality, but its resolution is not trivial since the syntactic role of temporal expression is understandable in the context. In our work, syntactic functions of temporal words are decisively identified based on local contexts of individual temporal words acquired from a large corpus, which are represented by a finite state method. Experimental results show how the proposed method, incorporated with parsing, improves the accuracy and efficiency of the syntactic analysis.

The choice of kernel is an important issue in the support vector machine algorithm, and the performance of it largely depends on the kernel. Up to now, no general rule is available as to which kernel should be used. In this paper we investigate two kernels: Gaussian RBF kernel and polynomial kernel. So far Gaussian RBF kernel is the best choice for practical applications. This paper shows that the polynomial kernel in the normalized feature space behaves better or as good as Gaussian RBF kernel. The polynomial kernel in the normalized feature space is the best alternative to Gaussian RBF kernel.

Radio frequency (RF) microwave can be used to predict glucose concentration in a sample. This paper presents preliminary results in determining the concentration by measuring relative permittivity in the solutions of distilled water, saline, human serum, and human blood containing glucose. In the experiments, sensitivity for detecting glucose concentration in blood solutions was almost 57 mg/dl at the frequency of approximately 5.8 GHz.