Subset Selection Scheme Research Articles

Energy Disaggregation via Learning Powerlets and Sparse Coding

In this paper, we consider the problem of energy disaggregation, i.e., decomposing a whole home electricity signal into its component appliances. We propose a new supervised algorithm, which in the learning stage, automatically extracts signature consumption patterns of each device by modeling the device as a mixture of dynamical systems. In order to extract signature consumption patterns of a device corresponding to its different modes of operation, we define appropriate dissimilarities between energy snippets of the device and use them in a subset selection scheme, which we generalize to deal with time-series data. We then form a dictionary that consists of extracted power signatures across all devices. We cast the disaggregation problem as an optimization over a representation in the learned dictionary and incorporate several novel priors such as device-sparsity, knowledge about devices that do or do not work together as well as temporal consistency of the disaggregated solution. Real experiments on a publicly available energy dataset demonstrate that our proposed algorithm achieves promising results for energy disaggregation.

Therapeutic index modeling and predictive QSAR of novel thiazolidin-4-one analogs against Toxoplasma gondii

The main idea of this study was to find predictive quantitative structure–activity relationships (QSAR) for the therapeutic index of 68 thiazolidin-4-one analogs against Toxoplasma gondii. Multivariate adaptive regression spline (MARS) together with Monte-Carlo (MC) sampling was proposed as a reliable descriptor subset selection strategy. Basis functions and knot points are also determined for each selected descriptor using generalized cross validation after frequency analysis. Least squares-support vector regression (LS-SVR) with optimized hyper-parameters was employed as mapping tool due to its promising empirical performance. The models were validated and tested through the use of the external prediction set of compounds, leave-one-out and leave-many-out cross validation methods, applicability domain analysis and Y-randomization. The robustness and accuracy of the QSAR models were confirmed by the satisfactory statistical parameters for the experimentally reported dataset (R2p=0.853, Q2LOO=0.785, R2L20%O=0.742 and r2m=0.715) and low standard error values (RMSEp=0.208, RMSELOO=0.321 and RMSEL20%O=0.376). The comprehensive analysis carried out in the present contribution using the proposed strategy can provide a considerable basis for the design and development of novel drug-like molecules against T.gondii.

Multibiometric System Using Level Set, Modified LBP and Random Forest

Multibiometric systems alleviate some of the shortcomings possessed by the unimodal biometrics and provide better recognition performance. This paper presents a multibiometric system that integrates the iris and face features based on the fusion at the feature level. The proposed multibiometric system has three novelties as compared to the previous works. First, distance regularized level-set evolution (DRLSE) technique is utilized to localize the iris and pupil boundary from an iris image. The DRLSE maintains the regularity of the level set function intrinsically during the curve evolution process and increases the numerical accuracy substantially. The proposed iris localization scheme is robust against poor localization and weak iris/sclera boundaries. Second, a modified local binary pattern (MLBP), which combines both the sign and magnitude features for the improvement of recognition performance, is applied. Third, to select the optimal subset of features from the fused feature vector, a feature subset selection scheme based on random forest (RF) is proposed. To evaluate the performance of the proposed scheme, the facial images of Yale Extended B Face database are fused with the iris images of CASIA V4 interval dataset to construct an iris–face multimodal biometric dataset. The experimental results indicate that the proposed multimodal biometrics system is more reliable and robust than the unimodal biometric scheme.

SLNC for Multi-Source Multi-Relay BICM-OFDM Systems

In this paper, we study the application of bit-interleaved coded modulation (BICM) and orthogonal frequency division multiplexing (OFDM) to reap the benefits of wireless multiuser network coding in practical frequency-selective fading channels. We propose a mapping based symbol level network coding (SLNC) scheme for a cooperative diversity system comprising multiple sources, multiple relays, and one common destination. A simple cooperative maximum-ratio combining scheme is used at the destination and is shown to successfully exploit both the full spatial and the full frequency diversity offered by the channel for arbitrary numbers of sources, arbitrary numbers of relays, and arbitrary linear modulation schemes. To gain analytical insight for system design, we derive a closed-form upper bound for the asymptotic worst-case pairwise error probability (PEP) and obtain the diversity gain of the considered SLNC scheme for BICM-OFDM systems. These analytical results reveal the influence of the various system parameters, such as the number of sources, the free distance of the code, and the frequency diversity of the involved links, on performance. Furthermore, we propose two different relay selection schemes for the considered system: a) bulk selection, i.e., a single best relay is selected to transmit on all sub-carriers, and b) per-subcarrier selection, where a best relay is selected on each sub-carrier. Last but not least, we exploit the derived PEP expression for selecting a subset of sources from the set of active sources when the number of active sources is larger than the number of available orthogonal relay channels. We study the achievable diversity gain for the proposed relay and source subset selection schemes. Numerical results corroborate the derived diversity gain expressions and confirm the performance gains.

Diagnosis of Breast Masses from Dynamic Contrast-Enhanced and Diffusion-Weighted MR: A Machine Learning Approach

PurposeDynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) is increasingly used for breast cancer diagnosis as supplementary to conventional imaging techniques. Combining of diffusion-weighted imaging (DWI) of morphology and kinetic features from DCE-MRI to improve the discrimination power of malignant from benign breast masses is rarely reported.Materials and MethodsThe study comprised of 234 female patients with 85 benign and 149 malignant lesions. Four distinct groups of features, coupling with pathological tests, were estimated to comprehensively characterize the pictorial properties of each lesion, which was obtained by a semi-automated segmentation method. Classical machine learning scheme including feature subset selection and various classification schemes were employed to build prognostic model, which served as a foundation for evaluating the combined effects of the multi-sided features for predicting of the types of lesions. Various measurements including cross validation and receiver operating characteristics were used to quantify the diagnostic performances of each feature as well as their combination.ResultsSeven features were all found to be statistically different between the malignant and the benign groups and their combination has achieved the highest classification accuracy. The seven features include one pathological variable of age, one morphological variable of slope, three texture features of entropy, inverse difference and information correlation, one kinetic feature of SER and one DWI feature of apparent diffusion coefficient (ADC). Together with the selected diagnostic features, various classical classification schemes were used to test their discrimination power through cross validation scheme. The averaged measurements of sensitivity, specificity, AUC and accuracy are 0.85, 0.89, 90.9% and 0.93, respectively.ConclusionMulti-sided variables which characterize the morphological, kinetic, pathological properties and DWI measurement of ADC can dramatically improve the discriminatory power of breast lesions.

BCG Artifact Removal for Reconstructing Full-scalp EEG inside the MR Scanner.

In simultaneous EEG/fMRI acquisition, the ballistocardiogram (BCG) artifact presents a major challenge for meaningful EEG signal interpretation and needs to be removed. This is very difficult, especially in continuous studies where BCG cannot be removed with averaging. In this study, we take advantage of a high-density EEG-cap and propose an integrated learning and inference approach to estimate the BCG contribution to the overall noisy recording. In particular, we present a special-designed experiment to enable a near-optimal subset selection scheme to identify a small set (20 out of 256 channels), and argue that in real-recording, BCG artifact signal from all channels can be estimated from this set. We call this new approach "Direct Recording Temporal Spatial Encoding" (DRTSE) to reflect these properties. In a preliminary evaluation, the DRTSE is combined with a direct subtraction and an optimization scheme to reconstruct the EEG signal. The performance was compared against the benchmark Optimal Basis Set (OBS) method. In the challenging nonevent-related EEG studies, the DRTSE method, with the optimization-based approach, yields an EEG reconstruction that reduces the normalized RMSE by approximately 13 folds, compared to OBS.

Active temperature modulation of metal-oxide sensors for quantitative analysis of gas mixtures

We present an active sensing method for quantitative analysis of gas mixtures using metal-oxide (MOX) chemical sensors. The method allows a MOX sensor to adapt its operating temperature in real time so as to sequentially reduce uncertainty in the concentration estimates. We formulate the problem as one of probabilistic state estimation coupled with a myopic optimization algorithm. At each iteration, the algorithm estimates the expected reduction in entropy for each sensing action (i.e., operating temperature) and selects the best such temperature. We first evaluated the proposed method on a simulated binary mixture problem using a computational model of MOX sensors. In these simulations, we compared the active sensing approach against conventional sequential forward selection (SFS) strategies. We then experimentally validated the method on a Taguchi gas sensor to quantify mixtures of two organic compounds. Our results indicate that the active sensing algorithm can obtain comparable estimation performance as SFS with significantly fewer measurements. In addition, since active sensing selects features on the fly, it is also more robust to experimental noise than off-line subset selection strategies.

Quality based classification of gasoline samples by ATR-FTIR spectrometry using spectral feature selection with quadratic discriminant analysis

A chemometric approach has been developed for characterization of gasoline samples regarding their quality. Attenuated total reflectance – infrared spectrometric data were processed by genetic algorithm (GA) and successive projection algorithm (SPA) feature selection techniques, being employed as an initial step prior to apply a discriminative tool. It was aimed to classify the fuel samples according to their quality passed/failed data. Chemometric predictive procedures were developed using quadratic discriminant analysis (QDA) combined with GA and SPA as a feature subset and feature selection strategy. Results showed 93.3% and 95.6% accuracy for SPA-QDA and GA-QDA models respectively.

On capacity gain of antenna subset selection for multiple antenna multicasting

AbstractIn this letter, we consider the transmit antenna subset selection scheme in multicast channels where the antenna subset is selected to maximise the minimum signal‐to‐noise power ratio among the associated users. Using an asymptotic analysis, the average capacity is derived for the large number of users asymptote under the assumption that the antennas are divided into non‐overlapping subsets, and each of them is independent, identical, spatially uncorrelated, and Rayleigh distributed. The numerical results demonstrate that the asymptotic average capacity is close to the exact capacity when signal‐to‐noise power ratio is not large or the number of users is large. Using this result, we compare the asymptotic average capacity of the selection scheme with that of the open‐loop multiple antenna multicasting scheme, in which only one antenna subset is employed, and its number of antennas is the same as that of the antenna subset of the selection scheme. Specifically, it is shown that the asymptotic capacity gain over the open‐loop scheme is significant when the number of antennas in each antenna subset is small. Copyright © 2013 John Wiley & Sons, Ltd.

A new hyper-heuristic as a general problem solver: an implementation in HyFlex

This study provides a new hyper-heuristic design using a learning-based heuristic selection mechanism together with an adaptive move acceptance criterion. The selection process was supported by an online heuristic subset selection strategy. In addition, a pairwise heuristic hybridization method was designed. The motivation behind building an intelligent selection hyper-heuristic using these adaptive hyper-heuristic sub-mechanisms is to facilitate generality. Therefore, the designed hyper-heuristic was tested on a number of problem domains defined in a high-level framework, i.e., HyFlex. The framework provides a set of problems with a number of instances as well as a group of low-level heuristics. Thus, it can be considered a good environment to measure the generality level of selection hyper-heuristics. The computational results demonstrated the generic performance of the proposed strategy in comparison with other tested hyper-heuristics composed of the sub-mechanisms from the literature. Moreover, the performance and behavior analysis conducted for the hyper-heuristic clearly showed its adaptive characteristics under different search conditions. The principles comprising the here presented algorithm were at the heart of the algorithm that won the first international cross-domain heuristic search competition.

Relay Subset Selection and Fair Power Allocation for Best and Partial Relay Selection in Generic Noise and Interference

Best relay selection (BRS) has received considerable attention in the literature as it makes efficient use of the system resources and achieves full diversity. Partial relay selection (PRS) is an alternative to BRS and performs relay selection based on local channel state information (CSI) only at the expense of a loss in diversity. Although, in practice, relays may be deployed in unfavorable environments exposing them to non-Gaussian impairments, existing analyses and design guidelines for BRS and PRS are limited to additive white Gaussian noise channels. In this paper, we analyze the error rate of BRS and PRS in the asymptotic regime of high signal-to-noise ratio (SNR) for amplify-and-forward (AF) relays and impairment by generic noise and interference. The derived analytical results are valid for Gaussian and non-Gaussian noises with finite moments, independent and non-identically distributed Rayleigh fading, and arbitrary linear modulation schemes. In order to reduce the signaling overhead required for CSI acquisition, we propose a relay subset selection scheme for BRS and PRS. Furthermore, to guarantee fairness in energy resource usage across the relays, we introduce a fair and flexible power allocation scheme with energy consumption constraints which does not affect the achievable diversity gain. The proposed relay subset selection and power allocation schemes only require knowledge of the average CSI of the links and certain moments of the noises impairing the relays and the destination.

QSPR-based estimation of the half-lives for polychlorinated biphenyl congeners

In this study, the depuration half-lives of 62 polychlorinated biphenyl (PCB) congeners in juvenile rainbow trout (Oncorhynchus mykiss) were estimated from their structural features based on QSPR methodology. A genetic algorithm (GA) was applied as a variable subset selection strategy and QSPR models established upon multiple linear regression (MLR), multilayer perceptron neural network (MLP NN) and support vector regression (SVR) procedures. Robustness and predictive stability of the constructed models were evaluated through internal and external validation methods. The high numerical values of and , and low RMSE in the case of the MLP NN model, confirm the supremacy of this model as well as nonlinear dependency of molecular structural features to the PCB congeners half-lives. In the best developed QSPR model the following four descriptors are used; lopping centric index (Lop), mean topological charge index of order 1 (JGI1), Geary autocorrelation lag-8/weighted by atomic Sanderson electronegativities (GATS8e) and highest eigenvalue of Burden matrix/weighted by atomic masses (BEHm3). Analysis of the descriptors involved in these models revealed that 2D molecular structural features, compactness and electronegativities are the main factors contributing to the half-lives of PCBs. The structural information presented in this work can be used for further evaluation of half-lives of PCBs and other similar structural compounds in the environment.

Automated Diagnosis of Alzheimer Disease using the Scale-Invariant Feature Transforms in Magnetic Resonance Images

In this paper we present an automated method for diagnosing Alzheimer disease (AD) from brain MR images. The approach uses the scale-invariant feature transforms (SIFT) extracted from different slices in MR images for both healthy subjects and subjects with Alzheimer disease. These features are then clustered in a group of features which they can be used to transform a full 3-dimensional image from a subject to a histogram of these features. A feature selection strategy was used to select those bins from these histograms that contribute most in classifying the two groups. This was done by ranking the features using the Fisher's discriminant ratio and a feature subset selection strategy using the genetic algorithm. These selected bins of the histograms are then used for the classification of healthy/patient subjects from MR images. Support vector machines with different kernels were applied to the data for the discrimination of the two groups, namely healthy subjects and patients diagnosed by AD. The results indicate that the proposed method can be used for diagnose of AD from MR images with the accuracy of %86 for the subjects aged from 60 to 80 years old and with mild AD.

Near-Optimal Relay Subset Selection for Two-Way Amplify-and-Forward MIMO Relaying Systems

This paper considers the relay subset selection problem in a two-way amplify-and-forward relay network, where each user is equipped with multiple antennas and each relay is equipped with a single antenna. In a two-way channel, two users exchange data with each other using a relay subset. With the optimal relay subset selection scheme, the system can reduce the hardware burden while preserving the diversity benefit. However, the optimal relay subset selection algorithm requires an exhaustive search of all possible combinations to determine the optimum subset in order to achieve the maximum sum-rate. Thus, this results in high computational complexity. In order to reduce the computational load while still maximizing the achievable rate, a cross-entropy (CE) method is introduced to search for the optimal relay subset. Simulation results indicate that the proposed CE method can guarantee a result that is within 98% of the optimum sum-rate obtained by the exhaustive search method with low computational complexity.

A Two-Stage Antenna Subset Selection Scheme for Amplify-and-Forward MIMO Relay Systems

A two-stage antenna subset selection scheme is proposed to maximize the channel capacity of amplify-and-forward multiple-input multiple-output relay systems. To reduce the computational complexity of the antenna selection process, the receive and transmit antenna selections are separated in a two-stage algorithm. In stage 1, a subset of receive antennas is selected at the relay node. A subset of transmit antennas is then selected at the relay node in stage 2. The simulation results demonstrate that the proposed scheme achieves nearly the same channel capacity with a reduced amount of computational complexity as compared to an exhaustive search.

Beam Subset Selection Strategy for Interference Reduction in Two-Tier Femtocell Networks

This paper examines an orthogonal random beamforming-based cross-tier interference reduction scheme for two-tier femtocell networks. In order to improve the immunity of both macrocell and femtocell users to cross-tier interference, we adopt a macrocell beam subset selection strategy. This beam subset selection strategy maximizes the throughput of the macrocell by optimizing the trade-off between the multiplexing gain and the multiuser interference. Simultaneously, the max-throughput scheduler suppresses the cross-tier interference with an adaptively reduced number of beams. Since the average cross-tier interference from the macrocell gradually decreases as the number of beams decreases, the beam subset selection strategy is capable of providing a spatial opportunity to the femtocell network. This spatial opportunity enables the femtocell network to take advantage of the selectivity arising from the correlation between the interference channel and the precoding matrix. Therefore, we also propose opportunistic channel selection and distributed power control strategies for the femtocell network. Both analytical and numerical results show that the proposed strategies collaboratively reduce the cross-tier interference in two-tier femtocell networks.

Dual-mode SM/STBC system with antenna subset selection employing ML detection

Multiple input multiple output (MIMO) antenna system is a promising candidate to meet the demands of 4th Generation (4G) cellular communication systems by offering increased spectral efficiency through the spatial multiplexing (SM) gain, and improved link reliability through the space–time block coding (STBC) diversity gain. This paper presents a new scheme that combines the dual-mode SM/STBC and the antenna subset selection (AnSS) schemes. In the proposed scheme, the combination of the SM/STBC switching and the full antenna subset selection (AnSS) at both the transmitter (Tx) and the receiver (Rx) ends of the communication channel are adaptively selected through a simple algorithm based on the singular values of the channel matrix at the Rx side. Thus, the new scheme achieves the best BER performance over the previous works regardless of the data rate. The simulation results show that the proposed scheme with the full AnSS outperforms the previous works, by up to the 12.5 dB at the bit error rate (BER) of 10 ‐ 5 . Further, a partial AnSS is also proposed which dramatically reduces both the computational complexity (by 31%) and the hardware (by 50%), cost, without any appreciable loss in the BER performance, when compared with the full AnSS.

Outage performance of multi-antenna multicasting for wireless networks

Wireless cellular networks often need to convey the same data to multiple users simultaneously. This kind of transmission is known as physical layer multicasting. Unfortunately, when any user fails to receive the data correctly, the information must be retransmitted to all the users, resulting in a waste of radio resources. Thus, it is important to investigate the outage probability of multicast channels, which is defined as the probability that the smallest maximum achievable rate among all of the users is smaller than a specified transmission rate. In this paper, we consider the use of multiple antenna multicast channels where the transmitter is equipped with M <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">t</sub> antennas and independent data is transmitted on each antenna to K users. Using extreme value theory, we derive a closed form limiting distribution that the exact distribution of the multicasting channel converges to when the number of users K is taken to infinity. From this result, we find upper and lower-bounds on the outage probability. It is shown that for a given outage probability the upper-bound becomes sufficiently tight to approximate the exact outage probability with K such that it converges to the exact one with a speed faster than Theta (K <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">-1/M</sup> <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">t</sup> ). Using this upper-bound, we identify conditions on the transmission rate and the transmit power necessary to maintain constant outage performance as K increases. Specifically, the transmission rate should be decreased as Theta (K <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">-1/M</sup> <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">t</sup> ) or the transmit power should be increased as Theta (K <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">1/M</sup> <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">t</sup> ). This means that the outage performance improves as the number of transmit antennas increases, because spatial diversity can be well exploited. However, this increase in the number of transmit antennas requires a significant cost. To improve the performance without requiring additional cost, we consider a multiple-slot multicasting that transmits the same data over multiple slots and an antenna subset selection scheme that transmits the data over some subset of the transmit antennas. It is shown that the transmission rate of multiple-slot scheme can be increased when the number of slots is carefully chosen and the diversity-multiplexing trade-off is the same as that of K = 1. Also, the gain in the transmission rate from selecting some subset of the transmit antennas is derived.

An Iterative Relative Entropy Minimization-Based Data Selection Approach for n-Gram Model Adaptation

Performance of statistical n-gram language models depends heavily on the amount of training text material and the degree to which the training text matches the domain of interest. The language modeling community is showing a growing interest in using large collections of text (obtainable, for example, from a diverse set of resources on the Internet) to supplement sparse in-domain resources. However, in most cases the style and content of the text harvested from the web differs significantly from the specific nature of these domains. In this paper, we present a relative entropy based method to select subsets of sentences whose n-gram distribution matches the domain of interest. We present results on language model adaptation using two speech recognition tasks: a medium vocabulary medical domain doctor-patient dialog system and a large vocabulary transcription system for European parliamentary plenary speeches (EPPS). We show that the proposed subset selection scheme leads to performance improvements over state of the art speech recognition systems in terms of both speech recognition word error rate (WER) and language model perplexity (PPL).

Space-Time Block Codes with Limited Feedback Using Antenna Grouping

We propose an antenna grouping method that improves the error rate performance of space-time codes in a wide range of mobility environments. The idea is to group symbols to antennas based on limited feedback from the mobile station to utilize all antennas. Our approach requires only two bits of feedback information to achieve better link performance and full rate for a certain four transmit antenna system. Numerical results confirm the bit/frame error gains over the Alamouti-based space-time block code and antenna subset selection strategies.