A new direction of arrival (DOA) estimation method is introduced with arbitrary array geometry when uncorrelated and coherent signals coexist. The DOAs of uncorrelated signals are first estimated via subspace-based high resolution DOA estimation technique. Then a matrix that only contains the information of coherent signals can be formulated by eliminating the contribution of uncorrelated signals. Finally a subspace block sparse reconstruction approach is taken for DOA estimations of the coherent signals.

We propose a committee-based method of active learning for large vocabulary continuous speech recognition. Multiple recognizers are trained in this approach, and the recognition results obtained from these are used for selecting utterances. Those utterances whose recognition results differ the most among recognizers are selected and transcribed. Progressive alignment and voting entropy are used to measure the degree of disagreement among recognizers on the recognition result. Our method was evaluated by using 191-hour speech data in the Corpus of Spontaneous Japanese. It proved to be significantly better than random selection. It only required 63 h of data to achieve a word accuracy of 74%, while standard training (i.e., random selection) required 103 h of data. It also proved to be significantly better than conventional uncertainty sampling using word posterior probabilities.

Previous studies of pattern recognition have shown that classifiers ensemble approaches can lead to better recognition results. In this paper, we apply the voting technique for the CoNLL-2010 shared task on detecting hedge cues and their scope in biomedical texts. Six machine learning-based systems are combined through three different voting schemes. We demonstrate the effectiveness of classifiers ensemble approaches and compare the performance of three different voting schemes for hedge cue and their scope detection. Experiments on the CoNLL-2010 evaluation data show that our best system achieves an F-score of 87.49% on hedge detection task and 60.87% on scope finding task respectively, which are significantly better than those of the previous systems.

Learning to rank refers to machine learning techniques for training the model in a ranking task. Learning to rank is useful for many applications in Information Retrieval, Natural Language Processing, and Data Mining. Intensive studies have been conducted on the problem and significant progress has been made [1],[2]. This short paper gives an introduction to learning to rank, and it specifically explains the fundamental problems, existing approaches, and future work of learning to rank. Several learning to rank methods using SVM techniques are described in details.

This letter presents a novel user scheduling algorithm that provides a maximum sum-rate based on zero-forcing beamforming (ZFBF) in multiple-input multiple-output (MIMO) systems. The proposed technique determines primary user pairs in which the sum-rate exceeds a predetermined threshold. To determine the threshold, we define the maximum-sum-rate criterion (MSRC) derived from the extreme value theory (EVT). Applying the MSRC in ZFBF-based user scheduling, we find that the performance of the proposed method is comparable to that of the exhaustive searching scheme which has a greater computational load. Through computer simulations, we show that the proposed method outperforms the very well-known correlation-based method, semi-orthogonal user selection (SUS), yielding a sum rate that is about 0.57 bps/Hz higher when the transmit SNR is 10 dB with perfect CSI at BS and the numbers of users and transmit antennas in a cell are 100 and 4, respectively.

In this paper, we show that the covariance-eigenvalue approach converges much faster than using cumulative distribution function (CDF) for determining diversity gain from channel measurements in reverberation chamber. The covariance-eigenvalue approach can be used for arbitrary multi-port antennas, but it is limited to Maximum Ratio Combining (MRC).

This paper proposes an efficient scheme for concurrent error detection for hardware implementations of the block cipher AES. In the proposed scheme, the circuit component for the round function is divided into two stages, which are used alternately for encryption (or decryption) and error checking in a pipeline. The proposed scheme has a limited overhead with respect to size and speed for the following reasons. Firstly, the need for a double number of clock cycles is eliminated by virtue of the reduced critical path. Secondly, the scheme only requires minimal additional circuitry for error detection since the detection is performed by the remaining encryption (or decryption) components within the pipeline. AES hardware with the proposed scheme was designed and synthesized by using 90-nm CMOS standard cell library with various constraints. As a result, the proposed circuit achieved 1.66 Gbps @ 12.9 Kgates for the compact version and 4.22 Gbps @ 30.7 Kgates for the high-speed version. These performance characteristics are comparable to those of a basic AES circuit without error detection, where the overhead of the proposed scheme is estimated to be 14.5% at maximum. The proposed circuit was fabricated in the form of a chip, and its error detection performance was evaluated through experiments. The chip was tested with respect to fault injection by using clock glitch, and the proposed scheme successfully detected and reacted to all introduced errors.

This paper proposes a novel waveguide intersection separating two H-plane waveguide systems from each other. If a four-port network in a four-fold rotational symmetry is completely matched, it has necessarily intersection properties. The proposed waveguide intersection consists of a square H-plane waveguide planar circuit connected four input/output waveguide ports in a four-fold rotational symmetry, and several metallic posts inserted at the junction without destroying the symmetry to realize a perfect matching. By optimizing the circuit parameters, high isolation properties are obtained in a relatively wide frequency band of about 8.6% for return loss and isolation better than 20 and 30 dB, respectively, for a circuit designed at 10 GHz. The proposed waveguide intersection can be analyzed by H-plane planar circuit approach, and possess advantages of compactness, simplicity, and high-power handling capability. Furthermore, an SIW intersection is designed by applying H-plane planar circuit approach to a waveguide circuit filled with dielectric material, and high isolation properties similar to H-plane waveguide intersection can be realized. The validity of these design concepts is confirmed by em-simulations and experiments.

Providing seamless QoS guarantees for multimedia services is one of the most critical requirements in the mobile Internet. However, the effects of host mobility make it difficult to provide such services. The next steps in signaling (NSIS) was proposed by the IETF as a new signaling protocol, but it fails to address some mobility issues. This paper proposes a new QoS NSIS signaling layer protocol (QoS NSLP) using a cross-layer design that supports mobility. Our approach is based on the advance discovery of a crossover node (CRN) located at the crossing point between a current and a new signaling path. The CRN then proactively reserves network resources along the new path that will be used after handoff. This proactive reservation significantly reduces the session reestablishment delay and resolves the related mobility issues in NSIS. Only a few amendments to the current NSIS protocol are needed to realize our approach. The experimental results and simulation study demonstrate that our approach considerably enhances the current NSIS in terms of QoS performance factors and network resource usage.

The contribution of this paper is two-fold. Firstly, we conduct a large-scale real-world evaluation of the effectiveness of integrating an automatic transliteration system with a machine translation system. A human evaluation is usually preferable to an automatic evaluation, and in the case of this evaluation especially so, since the common machine translation evaluation methods are affected by the length of the translations they are evaluating, often being biassed towards translations in terms of their length rather than the information they convey. We evaluate our transliteration system on data collected in field experiments conducted all over Japan. Our results conclusively show that using a transliteration system can improve machine translation quality when translating unknown words. Our second contribution is to propose a novel Bayesian model for unsupervised bilingual character sequence segmentation of corpora for transliteration. The system is based on a Dirichlet process model trained using Bayesian inference through blocked Gibbs sampling implemented using an efficient forward filtering/backward sampling dynamic programming algorithm. The Bayesian approach is able to overcome the overfitting problem inherent in maximum likelihood training. We demonstrate the effectiveness of our Bayesian segmentation by using it to build a translation model for a phrase-based statistical machine translation (SMT) system trained to perform transliteration by monotonic transduction from character sequence to character sequence. The Bayesian segmentation was used to construct a phrase-table and we compared the quality of this phrase-table to one generated in the usual manner by the state-of-the-art GIZA++ word alignment process used in combination with phrase extraction heuristics from the MOSES statistical machine translation system, by using both to perform transliteration generation within an identical framework. In our experiments on English-Japanese data from the NEWS2010 transliteration generation shared task, we used our technique to bilingually co-segment the training corpus. We then derived a phrase-table from the segmentation from the sample at the final iteration of the training procedure, and the resulting phrase-table was used to directly substitute for the phrase-table extracted by using GIZA++/MOSES. The phrase-table resulting from our Bayesian segmentation model was approximately 30% smaller than that produced by the SMT system's training procedure, and gave an increase in transliteration quality measured in terms of both word accuracy and F-score.

In this letter, a prerake combining scheme for signal detection in ultra-wideband (UWB) multiple input single output (MISO) systems with a hybrid pulse amplitude and position modulation (PAPM) is analytically examined. For a UWB MISO system, the analytical BER performance of a prerake combining scheme with PAPM is presented in a log-normal multipath fading channel. The analytical BERs are observed to match well the simulated results for the set of parameters chosen. The prerake diversity combining UWB systems, which can significantly reduce the complexity of the receiver side compared to the rake diversity systems, improve the error performance as the number of transmit antennas increases.

We propose a robust and efficient algorithm called ROCKET for clustering large-scale transaction databases. ROCKET is a divisive hierarchical algorithm that makes the most of recent hardware architecture. ROCKET handles the cases with the small and the large number of similar transaction pairs separately and efficiently. Through experiments, we show that ROCKET achieves high-quality clustering with a dramatic performance improvement.

A low-power low-noise intermediate-frequency (IF) circuit is proposed for Gaussian frequency shift keying (GFSK) low-IF receivers. The proposed IF circuit is realized by an all-analog architecture composed of a couple of limiting amplifiers (LAs) and received signal strength indicators (RSSIs), a couple of band-pass filters (BPFs), a frequency detector (FD), a low-pass filter (LPF) and a slicer. The LA and RSSI are realized by an optimized combination of folded amplifiers and current subtractor based rectifiers to avoid the process induced depressing on accuracy. In addition, taking into account the nonlinearity and static current of rectifiers, we propose an analytical model as an accurate approximation of RSSIs' transfer character. An active-RC based GFSK demodulation scheme is proposed, and then both low power consumption and a large dynamic range are obtained. The chip is implemented with HJTC 0.18 µm CMOS technology and measured under an intermediate frequency of 200 kHz, a data rate of 100 kb/s and a modulation index of 1. The RSSI has a dynamic range of 51 dB with a logarithmic linearity error of less than 1 dB, and the slope is 23.9 mV/dB. For 0.1% bit error ratio (BER), the proposed IF circuit has the minimum input signal-to-noise ratio (SNR) of 5 dB and an input dynamic range of 55.4 dB, whereas it can tolerate a frequency offset of -3%+9.5% at 6 dB input SNR. The total power consumption is 5.655.89 mW.

Adaptive network coded cooperation (ANCC) scheme may have excellent performance for data transmission from a large collection of terminals to a common destination in wireless networks. However, the random relay selection strategy for ANCC protocol may generate the distributed low-density parity-check (LDPC) codes with many short cycles which may cause error floor and performance degradation. In this paper, an optimized relay selection strategy for ANCC is proposed. Before data communication, by exploiting low-cost information interaction between the destination and terminals, the proposed method generates good assembles of distributed LDPC codes and its storage requirement reduces dramatically. Simulation results demonstrate that the proposed relay selection protocol significantly outperforms the random relay selection strategy.

The measure of the goodness, or inversely the cost, of concatenating synthesis units plays an important role in concatenative speech synthesis. In this paper, we present a probabilistic approach to concatenation modeling in which the goodness of concatenation is measured by the conditional probability of observing the spectral shape of the current candidate unit given the previous unit and the current phonetic context. This conditional probability is modeled by a conditional Gaussian density whose mean vector has a form of linear transform of the past spectral shape. Decision tree-based parameter tying is performed to achieve robust training that balances between model complexity and the amount of training data available. The concatenation models are implemented for a corpus-based speech synthesizer, and the effectiveness of the proposed method was confirmed by an objective evaluation as well as a subjective listening test. We also demonstrate that the proposed method generalizes some popular conventional methods in that those methods can be derived as the special cases of the proposed method.

This paper deals with underwater target classification using synthesized active sonar signals. Firstly, we synthesized active sonar returns from a 3D highlight model of underwater targets using the ray tracing algorithm. Then, we applied a multiaspect target classification scheme based on a hidden Markov model to classify them. For feature extraction from the synthesized sonar signals, a matching pursuit algorithm was used. The experimental results depending on the number of observations and signal-to-noise ratios are presented with our discussions.

We present an improvement to the existing steganography-based bandwidth extension scheme. Enhanced WB (wideband) speech quality is achieved by embedding multiple highband spectral gains into a G.711 bitstream. The number of spectral gains is selected by optimizing the quantity of the embedding data with respect to the quality of the extended WB speech. Compared to the existing method, the proposed scheme improves the WB PESQ (Perceptual Evaluation of Speech Quality) score by 0.334 with negligible degradation of the embedded narrowband speech.

Power analysis can be used to attack many implementations of cryptosystems, e.g., RSA and ECC, and the doubling attack is a collision based power analysis performed on two chosen ciphertexts. In this paper, we introduced a modified doubling attack to threaten RSA and ECC implementations by exploiting only one chosen ciphertext of small order. To attack the RSA implementations we selected an input of order two while to attack the ECC implementations we exploited one chosen invalid point of small order on a cryptographically weak curve rather than on the original curve. We showed that several existing power analysis countermeasures for RSA and ECC implementations are still vulnerable to the proposed attack. To prevent the proposed attack, we suggested countermeasures for RSA as well as for ECC.

This paper discusses recent developments for pattern recognition focusing on boosting approach in machine learning. The statistical properties such as Bayes risk consistency for several loss functions are discussed in a probabilistic framework. There are a number of loss functions proposed for different purposes and targets. A unified derivation is given by a generator function U which naturally defines entropy, divergence and loss function. The class of U-loss functions associates with the boosting learning algorithms for the loss minimization, which includes AdaBoost and LogitBoost as a twin generated from Kullback-Leibler divergence, and the (partial) area under the ROC curve. We expand boosting to unsupervised learning, typically density estimation employing U-loss function. Finally, a future perspective in machine learning is discussed.

In this paper, we present a distributed and interactive admission and power control protocol for spectrum underlay environments. The protocol enables distributed primary users (PUs) to estimate and adjust the level of tolerable interference as their transmitting powers evolve to a given signal-to-interference-plus-noise ratio (SINR) target. The protocol also guides the powers of distributed secondary users (SUs) to achieve their own targets while restricting the transmitting powers from SUs so as not to interfere with the PUs. This restriction of interference from SUs to PUs is an essential part of cognitive radio networks (CRNs) and is facilitated by sending a warning tone from PUs to SUs in the proposed protocol. The SUs that have frequently received the warning tones turn off their transmitters and so autonomously drop from the system. This paper proves that, under the proposed interactive protocol, every PU finally achieves its target if it is originally feasible without SUs and the transmit powers of remaining SUs converge to a fixed point. The proposed method protects PUs perfectly in the sense that all the PUs reach their targets after power control. Numerical investigation shows how safely PUs are protected and how well SUs are admitted as a function of protocol parameters, the frequency of warning tones, the number of SUs to be admitted and the number of active PUs.