This study proposes a variable selection linear regression (VSLR) adaptation framework to improve the accuracy of automatic speech recognition (ASR) with only limited and unlabeled adaptation data. The proposed framework can be divided into three phases. The first phase prepares multiple variable subsets by applying a ranking filter to the original regression variable set. The second phase determines the best variable subset based on a pre-determined performance evaluation criterion and computes a linear regression (LR) mapping function based on the determined subset. The third phase performs adaptation in either model or feature spaces. The three phases can select the optimal components and remove redundancies in the LR mapping function effectively and thus enable VSLR to provide satisfactory adaptation performance even with a very limited number of adaptation statistics. We formulate model space VSLR and feature space VSLR by integrating the VS techniques into the conventional LR adaptation systems. Experimental results on the Aurora-4 task show that model space VSLR and feature space VSLR, respectively, outperform standard maximum likelihood linear regression (MLLR) and feature space MLLR (fMLLR) and their extensions, with notable word error rate (WER) reductions in a per-utterance unsupervised adaptation manner.

Mobile Multi-hop Relay (MMR) technology is usually used to increase the transmission rate or to extend communication coverage. In this work, we show that MMR technology can also be used to raise the network capacity. Because Relay Stations (RS) are connected to the Base Station (BS) wirelessly and controlled by the BS, an MMR network can easily be deployed when necessary. High capacity MMR networks thus provide a good candidate solution for coping with temporary traffic surges. For the capacity enhancement of the MMR network, we suggest a novel scheme to parallelize cell transmissions while controlling the interference between transmissions. Using a numerical example for a typical network that is conformant to the IEEE 802.16j, we find that the network capacity increases by 88 percent.

Recent studies have focused on developing security systems using micro-Doppler radars to detect human bodies. However, the resolution of these conventional methods is unsuitable for identifying bodies and moreover, most of these conventional methods were designed for a solitary or sufficiently well-spaced targets. This paper proposes a solution to these problems with an image separation method for two closely spaced pedestrian targets. The proposed method first develops an image of the targets using ultra-wide-band (UWB) Doppler imaging radar. Next, the targets in the image are separated using a supervised learning-based separation method trained on a data set extracted using a range profile. We experimentally evaluated the performance of the image separation using some representative supervised separation methods and selected the most appropriate method. Finally, we reject false points caused by target interference based on the separation result. The experiment, assuming two pedestrians with a body separation of 0.44m, shows that our method accurately separates their images using a UWB Doppler radar with a nominal down-range resolution of 0.3m. We describe applications using various target positions, establish the performance, and derive optimal settings for our method.

Yield enhancements and quality improvements must be considered as factors of the utmost importance in VLSI (Very Large Scale Integration circuits) manufacturing in order to reduce cost and ensure customer satisfaction. This paper will present a study of the yield theory, an analysis of actual manufacturing data, and the challenges of yield enhancement.

This paper proposes the design of small CRPA arrays for dual-band Global Positioning System (GPS) applications. The array consists of five elements and is mounted on a circular ground platform with a diameter of 15-cm. Each antenna element has a coupled feed structure and consists of a feed patch and two radiating patches for dual-band operation. An external chip coupler is utilized for a broad circular polarization (CP) bandwidth, and its measured characteristics are taken into account in our simulation for more accurate performance estimation. Detailed parameters are optimized by using a genetic algorithm (GA) in conjunction with the FEKO EM simulator. The optimized antenna is fabricated on a ceramic substrate, and its performance is measured in a full anechoic chamber. Furthermore, a field test is also conducted to verify the signal-to-noise ratio (SNR) for real GPS satellite signals. The results prove that the proposed array is suitable for use in GPS CRPA applications.

Pattern-avoiding permutations are permutations where none of the subsequences matches the relative order of a given pattern. Pattern-avoiding permutations are related to practical and abstract mathematical problems and can provide simple representations for such problems. For example, some floorplans, which are used for optimizing very-large-scale integration (VLSI) circuit design, can be encoded into pattern-avoiding permutations. The generation of pattern-avoiding permutations is an important topic in efficient VLSI design and mathematical analysis of patten-avoiding permutations. In this paper, we present an algorithm for generating pattern-avoiding permutations, and extend this algorithm beyond classical patterns to generalized patterns with more restrictions. Our approach is based on the data structure πDDs, which can represent a permutation set compactly and has useful set operations. We demonstrate the efficiency of our algorithm by computational experiments.

In holographic data storage, information is recorded within the volume of a holographic medium. Typically, the data is presented as an array of pixels with modulation in amplitude and/or phase. In the 4-f orientation, the Fourier domain representation of the data array is produced optically, and this image is recorded. If the Fourier image contains large peaks, the recording material can saturate, which leads to errors in the read-out data array. In this paper, we present a coding process that produces sparse ternary data arrays. Ternary modulation is used because it inherently provides Fourier domain smoothing and allows more data to be stored per array in comparison to binary modulation. Sparse arrays contain fewer on-pixels than dense arrays, and thus contain less power overall, which reduces the severity of peaks in the Fourier domain. The coding process first converts binary data to a sequence of ternary symbols via a high-rate block code, and then uses guided scrambling to produce a set of candidate codewords, from which the most sparse is selected to complete the encoding process. Our analysis of the guided scrambling division and selection processes demonstrates that, with primitive scrambling polynomials, a sparsity greater than 1/3 is guaranteed for all encoded arrays, and that the probability of this worst-case sparsity decreases with increasing block size.

The transfer of prior knowledge from source domains can improve the performance of learning when the training data in a target domain are insufficient. In this paper we propose a new strategy to transfer deformable part models (DPMs) for object detection, using offline-trained auxiliary DPMs of similar categories as source models to improve the performance of the target object detector. A DPM presents an object by using a root filter and several part filters. We use these filters of the auxiliary DPMs as prior knowledge and adapt the filters to the target object. With a latent transfer learning method, appropriate local features are extracted for the transfer of part filters. Our experiments demonstrate that this strategy can lead to a detector superior to some state-of-the-art methods.

This paper describes a simple packet combining scheme with maximum likelihood detection (MLD) for multiple-input multiple-output with orthogonal frequency division multiplexing (MIMO-OFDM) in relay channels to construct reliable wireless links in wireless local area networks (LANs). Our MLD-based approach employs the multiplexed sub-stream signals in different transmit slots. The proposed scheme uses an additional combining process before MLD processing. Moreover, the proposed scheme sets the cyclic shift delay (CSD) operation in the relay terminal. We evaluate the performance of the proposed scheme by the packet error rate (PER) and throughput performance in the decode-and-forward (DF) strategy. First, we show that the proposed scheme offers approximately 4.5dB improvement over the conventional scheme in the received power ratio of the relay terminal to the destination terminal at PER =0.1. Second, the proposed scheme achieves about 1.6 times the throughput of the conventional scheme when the received power ratio of the relay terminal to the destination terminal is 3dB.

With the development of global navigation satellite systems (GNSS), the interference among global navigation satellite systems, known as the radio frequency compatibility problem, has become a matter of great concern to system providers and user communities. The acceptable compatibility threshold should be determined in the radio frequency compatibility assessment process. However, there is no common standard for the acceptable threshold in the radio frequency compatibility assessment. This paper firstly introduces the comprehensive radio frequency compatibility methodology combining the spectral separation coefficient (SSC) and code tracking spectral sensitivity coefficient (CT_SSC). Then, a method for determination of the acceptable compatibility threshold is proposed. The proposed method considers the receiver processing phase including acquisition, code and carrier tracking and data demodulation. Simulations accounting for the interference effects are carried out at each time step and every place on earth. The simulations mainly consider the signals of GPS, Galileo and BeiDou Navigation Satellite System (BDS) in the L1 band. Results show that all of the sole systems are compatible with other GNSS systems with respect to a special receiver configuration used in the simulations.

In this letter, we address monaural source separation based on supervised nonnegative matrix factorization (SNMF) and propose a new penalized SNMF. Conventional SNMF often degrades the separation performance owing to the basis-sharing problem. Our penalized SNMF forces nontarget bases to become different from the target bases, which increases the separated sound quality.

In this letter, we present a radix-R regular interconnection pattern family of factorizations for the WHT-FFT with identical stage-to-stage interconnection pattern in a unified form, where R is any power of 2. This family of algorithms has identical sparse matrix factorization in each stage and can be implemented in a merged butterfly structure, which conduce to regular and efficient memory managing scalable to high radices. And in each stage, the butterflies with same twiddle factor set are aggregated together, which can reduce the twiddle factor evaluations or accesses to the lookup table. The kinds of factorization can also be extended to FFT, WHT and SCHT with identical stage-to-stage interconnection pattern.

This letter presents a novel method based on sparsity, to solve the problem of deinterleaving pulse trains. The proposed method models the problem of deinterleaving pulse trains as an underdetermined system of linear equations. After determining the mixing matrix, we find sparsest solution of an underdetermined system of linear equations using basis pursuit denoising. This method is superior to previous ones in a number of aspects. First, spurious and missing pulses would not cause any performance reduction in the algorithm. Second, the algorithm works well despite the type of pulse repetition interval modulation that is used. Third, the proposed method is able to separate similar sources.

The millimeter-wave band suffers strong attenuation due to rain. While calculating the link budget for a wireless system using this frequency band, the behavior of rain, attenuation due to rain, and the amount of degradation must be accurately understood. This paper presents an evaluation of the influence of rain and its attenuation on link performance in a Tokyo Institute of Technology (Tokyo Tech) millimeter-wave model mesh network. Conventional statistical analyses including cumulative rain rate distribution and specific rain attenuation constants are performed on data collected from 2009 onwards. The unique effects arising due to the highly localized behaviors of strong rainfalls have become clear and are characterized in terms of variograms rather than correlation coefficients. Spatial separation even in the small network here with links of less than 1 km provides effective diversity branches for better availability performance.

The stochastic behavior of inter-core crosstalk in multi-core fiber is discussed based on a theoretical model validated by measurements, and the effect of the crosstalk on the Q-factor in transmission systems, using multi-core fiber is investigated theoretically. The measurements show that the crosstalk rapidly changes with wavelength, and gradually changes with time, in obedience to the Gaussian distribution in I-Q planes. Therefore, the behavior of the crosstalk as a noise may depend on the bandwidth of the signal light. If the bandwidth is adequately broad, the crosstalk may behave as a virtual additive white Gaussian noise on I-Q planes, and the Q-penalty at the Q-factor of 9.8dB is less than 1dB when the statistical mean of the crosstalk from other cores is less than -16.7dB for PDM-QPSK, -23.7dB for PDM-16QAM, and -29.9dB for PDM-64QAM. If the bandwidth is adequately narrow, the crosstalk may behave as virtually static coupling that changes very gradually with time and heavily depends on the wavelength. To cope with a static crosstalk much higher than its statistical mean, a margin of several decibels from the mean crosstalk may be necessary for suppressing Q-penalty in the case of adequately narrow bandwidth.

Collaborative business has been increasingly developing with the environment of globalization and advanced information technologies. In a collaboration environment with multiple organizations, participants from different organizations always have different views about modeling the overall business process due to different knowledge and cultural backgrounds. Moreover, flexible support, privacy preservation and process reuse are important issues that should be considered in business process management across organizational boundaries. This paper presents a novel approach of modeling interorganizational business process for collaboration. Our approach allows for modeling loosely coupled interorganizational business process considering different views of organizations. In the proposed model, organizations have their own local process views of modeling business process instead of sharing pre-defined global processes. During process cooperation, local process of an organization can be invisible to other organizations. Further, we propose the coordination mechanisms for different local process views to detect incompatibilities among organizations. We illustrate our proposed approach by a case study of interorganizational software development collaboration.

This paper describes a method for the efficient computation of the total autocorrelation for large multiple-output Boolean functions over a Shared Binary Decision Diagram (SBDD). The existing methods for computing the total autocorrelation over decision diagrams are restricted to single output functions and in the case of multiple-output functions require repeating the procedure k times where k is the number of outputs. The proposed method permits to perform the computation in a single traversal of SBDD. In that order, compared to standard BDD packages, we modified the way of traversing sub-diagrams in SBDD and introduced an additional memory function kept in the hash table for storing results of the computation of the autocorrelation between two subdiagrams in the SBDD. Due to that, the total amount of computations is reduced which makes the method feasible in practical applications. Experimental results over standard benchmarks confirm the efficiency of the method.

Developers often face difficulties while using APIs. API usage patterns can aid them in using APIs efficiently, which are extracted from source code stored in software repositories. Previous approaches have mined repositories to extract API usage patterns by simply applying data mining techniques to the collection of method invocations of API objects. In these approaches, respective functional roles of invoked methods within API objects are ignored. The functional role represents what type of purpose each method actually achieves, and a method has a specific predefined order of invocation in accordance with its role. Therefore, the simple application of conventional mining techniques fails to produce API usage patterns that are helpful for code completion. This paper proposes an improved approach that extracts API usage patterns at a higher abstraction level rather than directly mining the actual method invocations. It embraces a multilevel sequential mining technique and uses categorization of method invocations based on their functional roles. We have implemented a mining tool and an extended Eclipse's code completion facility with extracted API usage patterns. Evaluation results of this tool show that our approach improves existing code completion.

To better support data-intensive workflows which are typically built out of various independently developed executables, this paper proposes extensions to parallel database systems called User-Defined eXecutables (UDX) and collective queries. UDX facilitates the description of workflows by enabling seamless integrations of external executables into SQL statements without any efforts to write programs confirming to strict specifications of databases. A collective query is an SQL query whose results are distributed to multiple clients and then processed by them in parallel, using arbitrary UDX. It provides efficient parallelization of executables through the data transfer optimization algorithms that distribute query results to multiple clients, taking both communication cost and computational loads into account. We implement this concept in a system called ParaLite, a parallel database system based on a popular lightweight database SQLite. Our experiments show that ParaLite has several times higher performance over Hive for typical SQL tasks and has 10x speedup compared to a commercial DBMS for executables. In addition, this paper studies a real-world text processing workflow and builds it on top of ParaLite, Hadoop, Hive and general files. Our experiences indicate that ParaLite outperforms other systems in both productivity and performance for the workflow.

A suffix tree is widely adopted for indexing genome sequences. While supporting highly efficient search, the suffix tree has a few shortcomings such as very large size and very long construction time. In this paper, we propose a very fast parallel algorithm to construct a disk-based suffix tree for human genome sequences. Our algorithm constructs a suffix array for part of the suffixes in the human genome sequence and then converts it into a suffix tree very quickly. It outperformed the previous algorithms by Loh et al. and Barsky et al. by up to 2.09 and 3.04 times, respectively.