Seng KHEANG Kouichi KATSURADA Yurie IRIBE Tsuneo NITTA
The automatic transcription of out-of-vocabulary words into their corresponding phoneme strings has been widely adopted for speech synthesis and spoken-term detection systems. By combining various methods in order to meet the challenges of grapheme-to-phoneme (G2P) conversion, this paper proposes a phoneme transition network (PTN)-based architecture for G2P conversion. The proposed method first builds a confusion network using multiple phoneme-sequence hypotheses generated from several G2P methods. It then determines the best final-output phoneme from each block of phonemes in the generated network. Moreover, in order to extend the feasibility and improve the performance of the proposed PTN-based model, we introduce a novel use of right-to-left (reversed) grapheme-phoneme sequences along with grapheme-generation rules. Both techniques are helpful not only for minimizing the number of required methods or source models in the proposed architecture but also for increasing the number of phoneme-sequence hypotheses, without increasing the number of methods. Therefore, the techniques serve to minimize the risk from combining accurate and inaccurate methods that can readily decrease the performance of phoneme prediction. Evaluation results using various pronunciation dictionaries show that the proposed model, when trained using the reversed grapheme-phoneme sequences, often outperformed conventional left-to-right grapheme-phoneme sequences. In addition, the evaluation demonstrates that the proposed PTN-based method for G2P conversion is more accurate than all baseline approaches that were tested.
This paper claims to use a new question expansion method for question classification in cQA services. The input questions consist of only a question whereas training data do a pair of question and answer. Thus they cannot provide enough information for good classification in many cases. Since the answer is strongly associated with the input questions, we try to create a pseudo answer to expand each input question. Translation probabilities between questions and answers and a pseudo relevant feedback technique are used to generate the pseudo answer. As a result, we obtain the significant improved performances when two approaches are effectively combined.
Hideo FUJIWARA Katsuya FUJIWARA
In our previous work [12], [13], we introduced generalized feed-forward shift registers (GF2SR, for short) to apply them to secure and testable scan design. In this paper, we introduce another class of generalized shift registers called generalized feedback shift registers (GFSR, for short), and consider the properties of GFSR that are useful for secure scan design. We present how to control/observe GFSR to guarantee scan-in and scan-out operations that can be overlapped in the same way as the conventional scan testing. Testability and security of scan design using GFSR are considered. The cardinality of each class is clarified. We also present how to design strongly secure GFSR as well as GF2SR considered in [13].
Huifeng GUO Dianhui CHU Yunming YE Xutao LI Xixian FAN
Ranking as an important task in information systems has many applications, such as document/webpage retrieval, collaborative filtering and advertising. The last decade has witnessed a growing interest in the study of learning to rank as a means to leverage training information in a system. In this paper, we propose a new learning to rank method, i.e. BLM-Rank, which uses a linear function to score samples and models the pairwise preference of samples relying on their scores under a Bayesian framework. A stochastic gradient approach is adopted to maximize the posterior probability in BLM-Rank. For industrial practice, we have also implemented the proposed algorithm on Graphic Processing Unit (GPU). Experimental results on LETOR have demonstrated that the proposed BLM-Rank method outperforms the state-of-the-art methods, including RankSVM-Struct, RankBoost, AdaRank-NDCG, AdaRank-MAP and ListNet. Moreover, the results have shown that the GPU implementation of the BLM-Rank method is ten-to-eleven times faster than its CPU counterpart in the training phase, and one-to-four times faster in the testing phase.
Luis F. CISNEROS-SINENCIO Alejandro DIAZ-SANCHEZ Jaime RAMIREZ-ANGULO
Despite logic families based on floating-gate MOS (FGMOS) transistors achieve significant reductions in terms of power and transistor count, these logics have had little impact on VLSI design due to their sensitivity to noise. In order to attain robustness to this phenomenon, Positive-Feedback Floating-Gate logic (PFFGL) uses a differential architecture and positive feedback; data obtained from a 0.5µm ON Semiconductors test chip and from SPICE simulations shows PFFGL to be immune to noise from parasitic couplings as well as to leakage even when minimum device size is used.
To support the efficient gathering of diverse information about a news event, we focus on descriptions of named entities (persons, organizations, locations) in news articles. We extend the stakeholder mining proposed by Ogawa et al. and extract descriptions of named entities in articles. We propose three measures (difference in opinion, difference in details, and difference in factor coverage) to rank news articles on the basis of analyzing differences in descriptions of named entities. On the basis of these three measurements, we develop a news app on mobile devices to help users to acquire diverse reports for improving their understanding of the news. For the current article a user is reading, the proposed news app will rank and provide its related articles from different perspectives by the three ranking measurements. One of the notable features of our system is to consider the access history to provide the related news articles. In other words, we propose a context-aware re-ranking method for enhancing the diversity of news reports presented to users. We evaluate our three measurements and the re-ranking method with a crowdsourcing experiment and a user study, respectively.
Shanqi PANG Yajuan WANG Guangzhou CHEN Jiao DU
The orthogonal array is an important object in combinatorial design theory, and it is applied to many fields, such as computer science, coding theory and cryptography etc. This paper mainly studies the existence of the mixed orthogonal arrays of strength two with seven factors and presents some new constructions. Consequently, a few new mixed orthogonal arrays are obtained.
Lei CHEN Tapas Kumar MAITI Hidenori MIYAMOTO Mitiko MIURA-MATTAUSCH Hans Jürgen MATTAUSCH
In this paper, we report the design of an organic thin-film transistor (OTFT) driver circuit for the actuator of an organic fluid pump, which can be integrated in a portable-size fully-organic artificial lung. Compared to traditional pump designs, lightness, compactness and scalability are achieved by adopting a creative pumping mechanism with a completely organic-material-based system concept. The transportable fluid volume is verified to be flexibly adjustable, enabling on-demand controllability and scalability of the pump's fluid-flow rate. The simulations, based on an accurate surface-potential OTFT compact model, demonstrate that the necessary driving waveforms can be efficiently generated and adjusted to the actuator requirements. At the actuator-driving-circuit frequency of 0.98Hz, an all-organic fluid pump with 40cm length and 0.2cm height is able to achieve a flow rate of 0.847L/min, which satisfies the requirements for artificial-lung assist systems to a weakened normal lung.
Weijun ZENG Huali WANG Hui TIAN
In this letter, a new scheme for multirate coprime sampling and reconstructing of sparse multiband signals with very high carrier frequencies is proposed, where the locations of the signal bands are not known a priori. Simulation results show that the new scheme can simultaneously reduce both the number of sampling channels and the sampling rate for perfect reconstruction, compared to the existing schemes requiring high number of sampling channels or high sampling rate.
Carlos PEREZ-LEGUIZAMO P. Josue HERNANDEZ-TORRES J.S. Guadalupe GODINEZ-BORJA Victor TAPIA-TEC
Recently, the Services Oriented Architectures (SOA) have been recognized as the key to the integration and interoperability of different applications and systems that coexist in an organization. However, even though the use of SOA has increased, some applications are unable to use it. That is the case of mission critical information applications, whose requirements such as high reliability, non-stop operation, high flexibility and high performance are not satisfied by conventional SOA infrastructures. In this article we present a novel approach of combining SOA with Autonomous Decentralized Systems (ADS) in order to provide an infrastructure that can satisfy those requirements. We have named this infrastructure Autonomous Decentralized Service Oriented Architecture (ADSOA). We present the concept and architecture of ADSOA, as well as the Loosely Couple Delivery Transaction and Synchronization Technology for assuring the data consistency and high reliability of the application. Moreover, a real implementation and evaluation of the proposal in a mission critical information system, the Uniqueness Verifying Public Key Infrastructure (UV-PKI), is shown in order to prove its effectiveness.
Chia-Wen CHANG Kai-Yu LO Hossameldin A. IBRAHIM Ming-Chiuan SU Yuan-Hua CHU Shyh-Jye JOU
This paper presents a varactor-based all-digital phase-locked loop (ADPLL) with a multi-phase digitally controlled oscillator (DCO) for near-threshold voltage operation. In addition, a new all-digital reference spur suppression (RSS) circuit with multiple phases random-sampling techniques to effectively spread the reference clock frequency is proposed to randomize the synchronized DCO register behavior and reduce the reference spur. Because the equivalent reference clock frequency is reserved, the loop behavior is maintained. The area of the proposed spur suppression circuit is only 4.9% of the ADPLL (0.038 mm2). To work reliably at the near-threshold region, a multi-phase DCO with NMOS varactors is presented to acquire precise frequency resolution and high linearity. In the near-threshold region (VDD =0.52 V), the ADPLL only dissipates 269.9 μW at 100 MHz output frequency. It has a reference spur of -52.2 dBc at 100 MHz output clock frequency when the spur suppression circuit is deactivated. When the spur suppression circuit is activated, the ADPLL shows a reference spur of -57.3 dBc with the period jitter of 0.217% UI.
Yuta TAKATA Mitsuaki AKIYAMA Takeshi YAGI Takeo HARIU Shigeki GOTO
Drive-by download attacks force users to automatically download and install malware by redirecting them to malicious URLs that exploit vulnerabilities of the user's web browser. In addition, several evasion techniques, such as code obfuscation and environment-dependent redirection, are used in combination with drive-by download attacks to prevent detection. In environment-dependent redirection, attackers profile the information on the user's environment, such as the name and version of the browser and browser plugins, and launch a drive-by download attack on only certain targets by changing the destination URL. When malicious content detection and collection techniques, such as honeyclients, are used that do not match the specific environment of the attack target, they cannot detect the attack because they are not redirected. Therefore, it is necessary to improve analysis coverage while countering these adversarial evasion techniques. We propose a method for exhaustively analyzing JavaScript code relevant to redirections and extracting the destination URLs in the code. Our method facilitates the detection of attacks by extracting a large number of URLs while controlling the analysis overhead by excluding code not relevant to redirections. We implemented our method in a browser emulator called MINESPIDER that automatically extracts potential URLs from websites. We validated it by using communication data with malicious websites captured during a three-year period. The experimental results demonstrated that MINESPIDER extracted 30,000 new URLs from malicious websites in a few seconds that conventional methods missed.
Wei HAN Xiongwei ZHANG Gang MIN Meng SUN
In this letter, a novel perceptually motivated single channel speech enhancement approach based on Deep Neural Network (DNN) is presented. Taking into account the good masking properties of the human auditory system, a new DNN architecture is proposed to reduce the perceptual effect of the residual noise. This new DNN architecture is directly trained to learn a gain function which is used to estimate the power spectrum of clean speech and shape the spectrum of the residual noise at the same time. Experimental results demonstrate that the proposed perceptually motivated speech enhancement approach could achieve better objective speech quality when tested with TIMIT sentences corrupted by various types of noise, no matter whether the noise conditions are included in the training set or not.
Nattapong THAMMASAN Koichi MORIYAMA Ken-ichi FUKUI Masayuki NUMAO
Research on emotion recognition using electroencephalogram (EEG) of subjects listening to music has become more active in the past decade. However, previous works did not consider emotional oscillations within a single musical piece. In this research, we propose a continuous music-emotion recognition approach based on brainwave signals. While considering the subject-dependent and changing-over-time characteristics of emotion, our experiment included self-reporting and continuous emotion annotation in the arousal-valence space. Fractal dimension (FD) and power spectral density (PSD) approaches were adopted to extract informative features from raw EEG signals and then we applied emotion classification algorithms to discriminate binary classes of emotion. According to our experimental results, FD slightly outperformed PSD approach both in arousal and valence classification, and FD was found to have the higher correlation with emotion reports than PSD. In addition, continuous emotion recognition during music listening based on EEG was found to be an effective method for tracking emotional reporting oscillations and provides an opportunity to better understand human emotional processes.
Tomomi HATANO Takashi ISHIO Joji OKADA Yuji SAKATA Katsuro INOUE
For the maintenance of a business system, developers must understand the business rules implemented in the system. One type of business rules defines computational business rules; they represent how an output value of a feature is computed from the valid inputs. Unfortunately, understanding business rules is a tedious and error-prone activity. We propose a program-dependence analysis technique tailored to understanding computational business rules. Given a variable representing an output, the proposed technique extracts the conditional statements that may affect the computation of the output. To evaluate the usefulness of the technique, we conducted an experiment with eight developers in one company. The results confirm that the proposed technique enables developers to accurately identify conditional statements corresponding to computational business rules. Furthermore, we compare the number of conditional statements extracted by the proposed technique and program slicing. We conclude that the proposed technique, in general, is more effective than program slicing.
Bumsoon JANG Seokjoo DOO Soojin LEE Hyunsoo YOON
Due to the periodic recovery of virtual machines regardless of whether malicious intrusions exist, proactive recovery-based Intrusion Tolerant Systems (ITSs) are being considered for mission-critical applications. However, the virtual replicas can easily be exposed to attacks during their working period, and additionally, proactive recovery-based ITSs are ineffective in eliminating the vulnerability of exposure time, which is closely related to service availability. To address these problems, we propose a novel hybrid recovery-based ITS in this paper. The proposed method utilizes availability-driven recovery and dynamic cluster resizing. The availability-driven recovery method operates the recovery process by both proactive and reactive ways for the system to gain shorter exposure times and higher success rates. The dynamic cluster resizing method reduces the overhead of the system that occurs from dynamic workload fluctuations. The performance of the proposed ITS with various synthetic and real workloads using CloudSim showed that it guarantees higher availability and reliability of the system, even under malicious intrusions such as DDoS attacks.
Yuichi NAKAMURA Andy HARVATH Hiroaki NISHI
Changing attitudes toward energy security and energy conservation have led to the introduction of distributed power systems such as photovoltaic, gas-cogeneration, biomass, water, and wind power generators. The mass installation of distributed energy generators often causes instability in the voltage and frequency of the power grid. Moreover, the power quality of distributed power grids can become degraded when system faults or the activation of highly loaded machines cause rapid changes in power load. To avoid such problems and maintain an acceptable power quality, it is important to detect the source of these rapid changes. To address these issues, next-generation power grids that can detect the fault location have been proposed. Fault location demands accurate time synchronization. Conventional techniques use the Global Positioning System (GPS) and/or IEEE 1588v2 for time synchronization. However, both methods have drawbacks — GPS cannot be used in indoor situations, and the installation cost of IEEE 1588v2 devices is high. In this paper, a time synchronization technique using the broadcast function of an Ethernet Passive Optical Network (EPON) system is proposed. Experiments show that the proposed technique is low-cost and useful for smart grid applications that use time synchronization in EPON-based next-generation power grids.
Shoichiro YAMASAKI Tomoko K. MATSUSHIMA
Secret sharing is a method of information protection for security. The information is divided into n shares and reconstructed from any k shares, but no knowledge of the information is revealed from k-1 shares. Physical layer security is a method of achieving favorable reception conditions at the destination terminal in wireless communications. In this study, we propose a security enhancement technique for wireless packet communications. The technique uses secret sharing and physical layer security to exchange a secret encryption key. The encryption key for packet information is set as the secret information in secret sharing, and the secret information is divided into n shares. Each share is located in the packet header. The base station transmits the packets to the destination terminal by using physical layer security based on precoded multi-antenna transmission. With this transmission scheme, the destination terminal can receive more than k shares without error and perfectly recover the secret information. In addition, an eavesdropper terminal can receive less than k-1 shares without error and recover no secret information. In this paper, we propose a protection technique using secret sharing based on systematic Reed-Solomon codes. The technique establishes an advantageous condition for the destination terminal to recover the secret information. The evaluation results by numerical analysis and computer simulation show the validity of the proposed technique.
Asako SOGA Bin UMINO Yuho YAZAKI Motoko HIRAYAMA
This paper reports an assessment of the feasibility and the practicality of a creation support system for contemporary dance e-learning. We developed a Body-part Motion Synthesis System (BMSS) that allows users to create choreographies by synthesizing body-part motions to increase the effect of learning contemporary dance choreography. Short created choreographies can be displayed as animation using 3DCG characters. The system targets students who are studying contemporary dance and is designed to promote the discovery learning of contemporary dance. We conducted a series of evaluation experiments for creating contemporary dance choreographies to verify the learning effectiveness of our system as a support system for discovery learning. As a consequence of experiments with 26 students who created contemporary dances, we verified that BMSS is a helpful creation training tool to discover new choreographic methods, new dance movements, and new awareness of their bodies.
Arata KAWAMURA Noboru HAYASAKA Naoto SASAOKA
We propose an impact and high-pitch noise-suppression method based on spectral entropy. Spectral entropy takes a large value for flat spectral amplitude and a small value for spectra with several lines. We model the impact noise as a flat spectral signal and its damped oscillation as a high-pitch periodic signal consisting of spectra with several lines. We discriminate between the current noise situations by using spectral entropy and adaptively change the noise-suppression parameters used in a zero phase-based impact-noise-suppression method. Simulation results show that the proposed method can improve the perceptual evaluation of the speech quality and speech-recognition rate compared to conventional methods.