Search

Shannon’s entropy or an extension of Shannon’s entropy can be used to quantify information transmission between or among variables. Mutual information is the pair-wise information that captures nonlinear relationships between variables. It is more robust than linear correlation methods. Beyond mutual information, two generalizations are defined for multivariate distributions: interaction information or co-information and total correlation or multi-mutual information. In comparison to mutual information, interaction information and total correlation are underutilized and poorly studied in applied neuroscience research. Quantifying information flow between brain regions is not explicitly explained in neuroscience by interaction information and total correlation. This article aims to clarify the distinctions between the neuroscience concepts of mutual information, interaction information, and total correlation. Additionally, we proposed a novel method for determining the interaction information between three variables using total correlation and conditional mutual information. On the other hand, how to apply it properly in practical situations. We supplied both simulation experiments and real neural studies to estimate functional connectivity in the brain with the above three higher-order information-theoretic approaches. In order to capture redundancy information for multivariate variables, we discovered that interaction information and total correlation were both robust, and it could be able to capture both well-known and yet-to-be-discovered functional brain connections.

Vacating room after encryption (VRAE) is a popular framework of reversible data hiding for encrypted images (RDHEI). Most VRAE based RDHEI methods do not make a desirable payload. To address this issue, this paper proposes a novel data hiding technique using adaptive difference recovery (ADR) and exploits this novel technique to design an efficient VRAE based RDHEI method. Specifically, the proposed ADR-based data hiding technique adaptively determines the range of original difference between a cover pixel and a reference pixel with correlation according to the marked difference generated by bit substitution technique. The cover pixel can be recovered losslessly with the range of original difference and the reference pixel. In the proposed RDHEI method, the encrypted image is generated by block permutation and block-based modulation so that spatial redundancy within encrypted image blocks is preserved. Then the ADR-based data hiding technique is applied to the encrypted blocks. Experiment results show that the proposed RDHEI method outperforms some state-of-the-art RDHEI methods in payload. The average embedding rates of the proposed RDHEI method on the BOSSbase, BOWS-2 and UCID datasets are 3.2045, 3.1145 and 2.4633 bpp, respectively.

Feature selection based on entropy structure can be roughly divided into two types according to whether they are related to kernel density estimation (KDE). The first type is feature selection based on non-KDE entropy. The second type is feature selection based on KDE entropy. Compared with the first type, the second type avoids discretization and obtains more accurate mutual information when handling continuous data. However, existing feature selection methods based on KDE entropy neglect the fact that samples with noise have negative impacts on KDE. Besides, the feature evaluation functions don’t effectively assess relevance and redundancy of features. Thus, a feature selection method via maximizing independent and minimizing redundant classification information ratio is constructed based on adaptive weighted kernel density estimation. Firstly, an adaptive weighted kernel density estimation model is designed. Secondly, the entropy structure is defined by the adaptive weighted kernel density estimation model, and their theoretical properties are explored. Thirdly, a feature selection algorithm via maximizing independent and minimizing redundant classification information ratio is designed from the viewpoint of adaptive weighted kernel density estimation. Finally, comprehensive experiments are performed. The results illustrate that our approach has certain robustness, validity and superiority compared with other representative feature selection approaches.

Light field (LF) imaging has generated considerable interest owing to its ability to capture both spatial and angular information of light rays simultaneously. However, the extremely large volume of data associated with LF imaging poses challenges to both data storage and transmission. This study addresses this issue by proposing a view synthesis-based LF image compression method using a generative adversarial network (GAN). The primary basis of compression relies on the fact that adjacent sub-aperture images (SAIs) are highly correlated. Accordingly, only sparsely sampled SAIs are transmitted and the others are reconstructed at the decoder side. The proposed sparse SAI sampling method enhances the quality of reconstructed SAIs by considering a fair trade-off between the number of SAIs available for use as priors in the synthesis process and SAI redundancy. The quality of reconstructed SAIs is further enhanced by a GAN-based SAI synthesis method, where the synthesis procedure is broken into disparity estimation and un-sampled SAI estimation components, and the adversarial nature of the jointly trained generative and discriminative networks results in a more accurate generative model. Furthermore, more texture details can be preserved in the synthesized SAIs by adopting a loss function in the GAN model based on perceptual quality. Extensive experimental results demonstrate the superiority of the proposed method relative to several other state-of-the-art compression methods in terms of standard quality metrics and the perceptual quality of the synthetic SAIs at the decoder side.

Recently, a reversible data hiding scheme by redundant space transfer has been proposed by Liu et al. In this scheme, the encryption process consists of bit plane disordering, sub-block position disordering and different-time Arnold transform in each sub-block. Then, the additional data is embedded into the LSB planes of the encrypted image by compressing the redundant space. However, to transfer the redundancy of the plain image into encrypted image, this scheme essentially just performs three types of disorder operations, which is vulnerable to security analysis. In this paper, we firstly demonstrate that the encryption method of this scheme can be broken by chosen-plaintext attack (CPA). Furthermore, in order to achieve the resistance against CPA, we introduce 2-D Logistic-adjusted-sine map (2D-LASM) to produce the encryption sequences with the control parameters generated from the plain image. Apart from that, we also improve the sparse matrix compression to achieve a superior embedding capacity. Extensive experiments and analyses demonstrate that the proposed scheme outperforms the original one in terms of both security and embedding capacity.

Communication plays a vital role in world. It is mainly carried out in from of speech or through non-verbal movements. Though the communication through speech is the origin for communication, other body expressions such hand movement and facial expressions also help to communicate our thoughts to others. The main aim of this research is as follows; The main objective of image compression is to reduce the size of data i.e used to represent a digital media in an efficient manner without causing major degradation to the quality of the image. To increase the efficiency and performance of digital data, image compression is used in various fields like security industries, federal government agencies and health industries. The reduced size may facilitate the quick transfer of the data over the communication channels. There are two types of compression methods. One is lossy and second is lossless compression. In lossless compression, no information is lost that is reduces bits by finding and eliminating redundancy. Lossy compression technique reduces size of data by finding non-essential information and destroys it. Therefore, the process of decreasing the size of data file is termed as data compression.

Symbolic Regression searches for a mathematical expression that fits the input data set by minimizing the approximation error. The search space explored by this technique is composed of any mathematical function representable as an expression tree. This provides more flexibility for fitting the data but it also makes the task more challenging. The search space induced by this representation becomes filled with redundancy and ruggedness, sometimes requiring a higher computational budget in order to achieve good results. Recently, a new representation for Symbolic Regression was proposed, called Interaction-Transformation, which can represent function forms as a composition of interactions between predictors and the application of a single transformation function. In this work, we show how this representation can be modeled as a multi-layer neural network with the weights adjusted following the Extreme Learning Machine procedure. The results show that this approach is capable of finding equally good or better results than the current state-of-the-art with a smaller computational cost.

Pole number recognition is highly important for the positioning tasks in high-speed rail catenary systems. The complicated working environment poses difficulties for number recognition algorithms, and this situation becomes even more challenging when illumination changes, image blurs and occlusions are considered. In this work, we present a high-accuracy pole number recognition framework including a high-performance cascaded CNN-based Detection and Recognition YOLO (DR-YOLO) and a temporal redundancy approach. First, the DR-YOLO utilizes global features for coarse plate detection and local features for accurate number recognition. Next, context-based combination of adjacent frames utilizes the complementarity and consistency of the same pole number recognition results and generates a unique result. The context-based amendment of adjacent poles utilizes the continuity of adjacent poles and corrects the fault or partial missing results caused by blur or occlusions. Extensive experimental testing is performed on 4 datasets representing high-speed train routine working environments. The reported experimental results validate the effectiveness and efficiency of the proposed approach.

Spectral analysis and Information-theory are two powerful and successful frameworks for feature selection in supervised classification problems. However, most of the methods developed under these frameworks have been introduced for handling exclusively numerical or non- numerical data. In this paper, we propose a supervised filter feature selection method that combines Spectral Feature Selection and Information-theory based redundancy analysis for selecting relevant and non-redundant features in supervised mixed datasets; i.e., datasets where the objects are described simultaneously by both, numerical and non-numerical features. To demonstrate the effectiveness of our proposed supervised filter feature selection method, we conducted several experiments on 40 public real-world datasets. Additionally, we compare our method against relevant state-of-the-art supervised filter methods for numerical, non-numerical, and mixed data. From this comparison, our method, in general, obtains better results than the results obtained by the other evaluated filter feature selection methods.

Vision is the main way for people to perceive and recognize the world. In this paper, four categories of the redundant information of video encoding, spatial redundancy, time redundancy, visual redundancy and statistical redundancy, are first introduced. Spatial redundancy is redundant information existing in static frames, time redundancy is the luminance and chrominance correlation between adjacent frames in the video sequence, visual redundancy is information that the human visual system cannot perceive, statistical redundancy is the expression redundancy of the information source symbols. Then the performance assessment of video coding are introduced. And the future development of video coding and conclusion are finally discussed in this paper.

Intra prediction is an effective method for video coding to remove the spatial redundancy of content. Classical intra prediction method usually creates a prediction block by extrapolating the encoded pixels surrounding the target block. However, existing methods cannot guarantee the prediction efficiency for rich textural structure, especially when weak spatial correlation exists between the target block and reference pixels. To remedy this issue, this paper proposes a novel intra prediction method via convolutional encoder-decoder network, which we term IPCED. IPCED can learn and extract the internal representation of reference blocks, and progressively generate a prediction block from this representation. IPCED is a data-driven method, which represents an improvement over hand-crafted methods, and is capable of improving the accuracy of intra prediction. Extensive experimental results demonstrate that IPCED can generate higher-quality intra prediction results, achieves 3.41%, 3.07% and 3.44% bitrate saving for the Y/Cb/Cr channel compared with HEVC baseline, which is significantly beyond existing methods.

In this letter, we propose a concise feature representation framework for acoustic scene classification by pruning embeddings obtained from SoundNet, a deep convolutional neural network. We demonstrate that the feature maps generated at various layers of SoundNet have redundancy. The proposed singular value decomposition based method reduces the redundancy while relying on the assumption that useful feature maps produced by different classes lie along different directions. This leads to ignoring the feature maps that produce similar embeddings for different classes. In the context of using an ensemble of classifiers on the various layers of SoundNet, pruning the redundant feature maps leads to reduction in dimensionality and computational complexity. Our experiments on acoustic scene classification demonstrate that ignoring 73% of feature maps reduces the performance by less than 1% with 12.67% reduction in computational complexity. In addition to this, we also show that the proposed pruning framework can be utilized to remove filters in the SoundNet network architecture, with 13x lesser model storage requirement. Also, the number of parameters reduce from 28 million to 2 million with marginal degradation in performance. This small model obtained after applying the proposed pruning procedure is evaluated on different acoustic scene classification datasets, and shows excellent generalization ability.

In category-theoretic models for the anyon systems proposed for topological quantum computing, the essential ingredients are two monoidal structures, $\oplus$ and $\otimes$. The former is symmetric but the latter is only braided, and $\otimes$ is required to distribute over $\oplus$. What are the appropriate coherence conditions for the distributivity isomorphisms? We came to this question working on a simplification of the category-theoretical foundation of topological quantum computing, which is the intended application of the research reported here. This question was answered by Laplaza when both monoidal structures are symmetric, but topological quantum computation depends crucially on $\otimes$ being only braided, not symmetric. We propose coherence conditions for distributivity in this situation, and we prove that our conditions are (a) strong enough to imply Laplaza's when the latter are suitably formulated, and (b) weak enough to hold when --- as in the categories used to model anyons --- the additive structure is that of an abelian category and the braided $\otimes$ is additive. Working on these results, we found a new redundancy in Laplaza's conditions., This is a companion paper for article "Witness algebra and anyon braiding," arXiv:1807.10414, proving results used there

The real world data are complex in nature and addition to that a large number of features add more value to the complexity. However, the features associated with the data may be redundant and erroneous in nature. To deal with such type of features, feature selection plays a vital role in computational learning. The reduction in the dimensionality of the dataset not only reduces the computational time required for classification but also enhances the classification accuracy by removing the misleading features. This paper presents a Filter Approach using Elitism based Multi-objective Differential Evolution algorithm for feature selection (FAEMODE) and the novelty lies in the objective formulation, where both linear and nonlinear dependency among features have been considered to handle the redundant and unwanted features of a dataset. Finally, the selected feature subsets of 23 benchmark datasets are tested using 10-fold cross validation with four well-known classifiers to endorse the result. A comparative analysis of the proposed approach with seven filter approaches and two conventional as well as three metaheuristic based wrapper approaches have been carried out for validation. The result reveals that the proposed approach can be considered as a powerful filter method for feature selection in various fields. Keywords: Multi-objective, Feature selection, Differential Evolution, Filter approach, Correlation coefficient, Mutual information

Because of the massive development of the Internet of Things (IoT), the quantity of remote devices associated with the Internet is estimated to touch 26 billion by 2020. This will challenge both the energy proficiency of wireless battery fuelled gadgets and the bandwidth capacities of wireless networks. One answer for both difficulties could be to use packet header compression. The primary reason for having header compression is that there is a huge amount of redundancy in the header fields of the packets owned by the same packet stream. Hence, there is a need to reduce this redundancy to make the best use of the available bandwidth so as to make the processing of data faster. This can be done by bringing the compressor and decompressor in sync with each other. To reduce the header size considerably, sending static/fixed information only initially followed by making predictions and the use of dependencies for rest of the fields fructifies. This paper reviews various existing header compression schemes, identifies the need for Robust Header Compression (ROHC) scheme and introduces the header compression of 6LoWPAN. 6LoWPAN provides header compression mechanisms to reduce the size of upper layer headers. 6LoWPAN header compression mechanisms can be used to compress the security headers as well.

Feature selection is a crucial step nowadays in machine learning and data analytics to remove irrelevant and redundant characteristics and thus to provide fast and reliable analyses. Many research works have focused on developing new methods that increase the global relevance of the subset of selected features while reducing the redundancy of information. However, those methods that select features with high relevance and low redundancy are extremely time-consuming when processing large datasets. In this work we present CUDA-JMI, a tool based on Joint Mutual Information that accelerates feature selection by exploiting the computational capabilities of modern heterogeneous systems that contain several CPU cores and GPU devices. The experimental evaluation has been carried out in three systems with different type and amount of CPUs and GPUs using five publicly available datasets from different fields. These results show that CUDA-JMI is significantly faster than its original sequential counterpart for all systems and input datasets. For instance, the runtime of CUDA-JMI is up to 52 times faster than an existing sequential JMI-based implementation in a machine with 24 CPU cores and two NVIDIA M60 boards (four GPUs). CUDA-JMI is publicly available to download from https://sourceforge.net/projects/cuda-jmi .

Multi-label feature selection plays an important role in pattern recognition, which can improve multi-label classification performance. In traditional multi-label feature selection methods based on information theory, feature relevance is evaluated by the accumulated mutual information between a candidate feature and each label. However, to the best of our knowledge, traditional methods ignore the effect of label redundancy on the evaluation of feature relevance. To address this issue, we propose a new multi-label feature selection method named multi-label Feature Selection based on Label Redundancy (LRFS). First, we categorize labels into two groups: independent labels and dependent labels. Second, by analyzing the differences between independent labels and dependent labels, we propose a new feature relevance term, that is, the conditional mutual information between candidate features and each label given other labels. Finally, we combine the new feature relevance term with the feature redundancy term to design our feature selection method. To evaluate the classification performance of our method, LRFS is compared to three information-theoretical-based multi-label feature selection methods on an artificial data set. Furthermore, LRFS is compared to five algorithm adaption feature selection methods and two problem transformation feature selection methods on 12 real-world multi-label data sets. The experimental results demonstrate that LRFS outperforms the other compared methods in terms of four evaluation metrics.

A trajectory optimization strategy is proposed in this paper to inhibit joint parameter jumps, which are unavoidable when a locked-joint failure occurs in a load-carrying task without shutting down for a space manipulator. By selecting objective functions and constraint conditions, a trajectory optimization model is established. To solve the model, global null-space correction terms containing compensation coefficients are introduced into the joint velocity and torque by utilizing the redundancy of the manipulator. Then by selecting the optimal compensation coefficients of all possible failure moments and fitting them as functions, which are used to correct the joint parameters during the whole task cycle, the inhibition of the joint parameter jump in a sudden failure is realized. Finally, simulations are carried out surrounding a 7-DOF space manipulator. It indicates the average inhibition ratios of the joint velocity and torque jump reach values of up to 95 and 93%, respectively, verifying the correctness and effectiveness of the trajectory optimization strategy. The strategy has two contributions: (1) both the joint velocity and torque jump can be optimally inhibited, and the average inhibition ratios reach 90% at least; (2) the trajectory optimization strategy is practical for a joint failure occurring at any moment.

This paper proposes a reversible data hiding scheme for encrypted images with SHE (Somewhat Homomorphic Encryption) based on sorting block-level prediction-error expansion (SBPEE). The image provider divides the image into 2 × 2 image blocks and uses SHE to protect the content of the image. In encrypted images, the data hider embeds the secret data with the method of SBPEE, which is based on the pixel redundancy within each block. And the sorting technique shows a good performance at low embedding rate. With the watermarked encrypted image, if the receiver has only the data hiding key, he can extract the additional data. If the receiver has both the data hiding key and the encryption key, he can extract the additional data correctly and recover the original image without loss. The experimental results show that the performance of the method is better than that of the existing similar methods when the embedding rate is low and is consistent with the existing methods when the embedding rate is high.

The existing reversible data hiding (RDH) methods based on improved pixel value ordering (IPVO) use only a small portion of the pixels (e.g., at most six pixels) in a block for embedding data. Our proposed method is a dynamic IPVO RDH, which can flexibly modify the number of pixels in a block by classifying the local complexity into multiple levels. For a highly-correlated, smooth block, almost all of the pixels, except for one or two in the middle positions, can be involved in embedding data. For a lowly-correlated smooth block, only two pixels can be embedded with data, and no data can be embedded into a highly-textured block. Thus, it is apparent that the redundancy among pixels is fully exploited, and further, more data can be embedded into smooth blocks while maintaining lower distortion. However, when all of the levels are used for embedding data, the visual quality may not be maximized for a given payload. Hence, in our method, we traverse from the first level to the highest level to search for the optimal number of levels that can provide the best embedding performance. Extensive experiments have demonstrated that the proposed method outperforms the existing methods.

Combination models are regarded as a popular approach to improve forecasting accuracy. Determining how to select optimal sub-models from all possible candidate models is considered to be a challenge in combination models because of the high computational overhead. However, there is fewer studies on sub-models selection for combination approaches. To this end, this study proposes a novel sub-models selection algorithm named neighborhood mutual information (NMI) theory-based maximizing linear relevance and minimizing linear redundancy (NMI-MRMR) to select the optimal sub-models set. Cuckoo search (CS) is used to find the optimal parameter of a neighborhood for the proposed algorithm by avoiding the evaluation of all possible parameters. Three cases are used to validate the performance of the proposed approach. The experimental forecasting results demonstrate that the combination of the individual models selected by the NMI-MRMR based CS algorithm significantly outperforms the combination of the individual models selected by traditional approaches. Furthermore, a comparison with two other artificial intelligence algorithms and a traversal algorithm indicate that the CS algorithm is more suitable for choosing the optimal parameter for the proposed algorithm. The proposed approach not only overcomes the shortcomings of combining all individual models but is also able to select the optimal sub-models set.

The sensor placement methods for the modal identification are usually based on the prediction error between the measured and the structural model responses. Generally, the prediction errors, which are caused by the model error and the measurement noise, are assumed to be a Gaussian random vector together. This paper proposes a conditional information entropy based sensor placement method to separately investigate the influences of the measurement noise and the model error on the sensor placement. The changes in the element stiffness matrices are used to represent the model error. To quantify the uncertainty in the modal identification result, the conventional information entropy method is not suitable, due to the uncertainty in the Fisher matrix caused by the uncertain model error. This paper proposes a conditional information entropy index to quantify the uncertainty in the modal identification result with uncertain Fisher matrices. The optimal sensor placement corresponds to the one that maximizes the conditional information entropy index value over the set of all possible sensor placements. The conventional information entropy method can be seen as a special case of the proposed conditional information entropy method. In addition, the proposed sensor placement method can be applied to the placement of the multidimensional sensors. The simply supported beam and a bridge benchmark structure are used to explain the sensor placement method, and the redundancy threshold can be set to avoid redundant information provided by closely spaced sensors.

Considering the limitation of selecting several neighbor sensors in a local region similar to just single one, namely redundant information, a sensor placement algorithm for structural health monitoring is proposed based on sub-clustering strategy, in order to improve the performance of sensor configuration with less redundancy. According to the significance of redundancy, the proposed novel redundancy elimination model considers global and local effect to overcome the previous limitations in sensor distribution. Based on the sub-clustering strategy, the redundancy elimination model reflects the sensor configuration in each sub-clustering and overall structural field. The presented sub-clustering strategy for sensor placement includes three main procedures: sub-clustering algorithm, its check step and smallest enclosing circle method, thus the accuracy can be guaranteed. Combining the effective independence method with normalization and weighting factor, the proposed sensor placement algorithm can balance performance and redundancy by using genetic algorithm, which is more competitive to reduce the order difference between the two objectives. Finally, the effectiveness of the proposed redundancy elimination model is verified by a simple example, and another two engineering numerical examples including space solar power satellite and re-usable launch vehicle are applied to demonstrate the validity of the proposed sensor placement algorithm respectively.

Image inpainting is a classical yet challenging inverse ill-posed problem. In this paper, we introduce the multi-filters guided low-rank tensor coding as a prior information to tackle it. The key innovation is to formulate multiple feature-domain tensors by convoluting the target image with multi-filters. Furthermore, by exploring a low-rank tensor coding, it can reduce the redundancy between sparse feature vectors at neighboring locations and improve the efficiency of the overall representation. The resulting non-convex model is iteratively tackled by gradient descent procedure for updating the image and by low-rank pursuit procedure for updating the multi-view features. Besides, we explore an aggregation version of proposed method for further improving the inpainting performance. Experimental results demonstrate that the proposed algorithms can faithfully recover image and outperform the current state-of-the-art approaches in terms of visual inspection, the quantitative peak signal-to-noise ratio (PSNR), and structural similarity (SSIM).

This paper presents a video coding scheme tailored for traffic surveillance videos, which features a pre-built library that is utilized in both encoder and decoder to pursue higher compression efficiency. We are motivated by the observation that, in traffic surveillance videos, not only the background is steady for a long while, but also the foreground (e.g. vehicles) contains high redundancy. For example, in the video taken by a traffic monitoring camera, we can observe that the passing-through vehicles are usually similar. However, the redundancy in the vehicles and the background is not fully exploited in the current video coding schemes. To address this problem, we propose a library-based video coding scheme. Specifically, for each static monitoring camera, we can collect video in a period, and build a library of vehicles and background from that video. When encoding the following video of that camera, we utilize the pre-built library by searching similar vehicles and background from the library and using the retrieved vehicles and background to help encode. Accordingly, the decoder also refers to the same library to reconstruct the video. We design efficient algorithms for building the library, searching in the library, as well as utilizing the library for encoding/decoding, to fulfill the proposed scheme. Our scheme is implemented upon the state-of-the-art video coding system–High Efficiency Video Coding (HEVC), and is tested on our own collected traffic surveillance videos. Experimental results show that, compared to the HEVC anchor, our proposed scheme achieves as high as 47.0%, 37.1%, and 34.8% BD-rate reduction, under random-access, low-delay B, and low-delay P settings, respectively. Our scheme also achieves higher compression efficiency than the existing background-based methods for surveillance video coding.

In LC-MS based untargeted analysis, data is collected at the peak or ion level, although the investigated biochemistry processes occur at the compound or reaction level. To this end, the presence of redundancy peaks such as co-eluted peaks, multi-chargers, adducts, neutral loss, isotopologues, and fragments ions often muddle subsequent statistical data analysis. In order to fill this gap, between peaks and compounds/reactions, independent components must first be found at the peak level, then evaluated at the compound or reaction levels. Based on paired mass distances (PMD), the algorithm GlobalStd, based on retention time hierarchical cluster analysis and global analysis of PMDs within clusters, is here proposed to extract independent peaks from raw LC-MS data. Following its application, a structure/reaction directed analysis can then be used to evaluate compounds at the structure or biochemistry reaction level, based on similar PMDs among different retention times clusters. As a proof-of-concept, the developed statistical method was applied to data obtained for in vivo SPME sampling on fish. In total, 277 independent peaks were demonstrated to stand for most of the variances found for the total 1459 ions detected via LC-MS. Following, both known homologous series or biological reactions along with unknown bio-processes, which may involve oxidation/reduction reactions or homologous series, were analyzed via a structure/reaction directed analysis. The findings of this analysis yielded interesting information regarding the data, for instance denoting the possible occurrence of a biosynthesis process involving l -Carnitine and its precursor 4-Trimethylammoniobutanoic acid. Such PMD relationships could also aid in the screening of annotation results. To this end, semi-quantitative analysis based on structure/reaction directed analysis is also here proposed for further investigation of unknown patterns or for removal of contaminants in metabolomics studies. The developed data-driven algorithm has been included in a PMD package with a GUI interactive document, and is freely available online ( https://github.com/yufree/pmd ).

Object proposal has been successfully applied in recent supervised and weakly supervised visual object detection tasks to improve the computational efficiency. The classical grouping-based object proposal approach can produce region proposals with high localization accuracy, but incorporates significant redundancy for the lack of object confidence to evaluate the proposals. In this paper, we propose leveraging the essential properties of images, i.e., contour and symmetry, to score the redundant region proposals. Specifically, the contour and symmetry are extracted by a Simultaneous Contour and Symmetry Detection Network (SCSDN) and used to score the bounding box with a Bayesian framework, which guarantees that the scoring procedure is adaptive to general objects. A subset of high-scored proposals reserves the recall rate, while can also significantly decrease the redundancy. Experimental results show that the proposed approach improves the baseline by increasing the recall rate from 0.87 to 0.89 on the PASCAL VOC 2007 dataset. It also outperforms the state-of-the-art on AUC and uses much fewer object proposals to achieve comparable recall rate.

PURPOSE/OBJECTIVE(S) Local failure (LF) following chemoradiation for head and neck squamous cell carcinoma (HNSCC) is associated with poor overall survival. If machine learning techniques could stratify patients at risk of treatment failure based on baseline and intra-treatment imaging, such a model could facilitate response-adapted approaches to escalate, de-escalate, or switch therapy. Our hypothesis was that adding baseline CT (CT1) and delta CBCT radiomic features to a clinical model would improve its discriminatory ability at predicting LF. MATERIALS/METHODS Participants treated at a single institution with definitive radiotherapy (RT) for HNSCC and who failed locally in-field at a primary or nodal structure (analyzed separately) were included and matched to controls (30 cases, 60 controls) based on the site of primary, T/N stage, and p16 status. Radiomic features were extracted from CT1 and CBCT scans at fractions 1 (CBCT01) and 21 (CBCT21; where CBCT21 - CBCT01 = delta features) of RT with open-source software and were selected for by: reproducibility (intra-class correlation coefficients < 0.95), redundancy (maximal relevance and minimal redundancy), and informativeness (recursive feature elimination). Separate models predicting LF of primaries or nodes were created using the explainable boosting machine classifier with 5-fold cross-validation for (1) clinical only, (2) radiomic only (CT1 and delta features), and (3) fused models (clinical and radiomic models fused). Twenty-five iterations were performed and predicted scores were averaged with a parallel ensemble design. Receiver operating characteristic (ROC) curves were compared against each other with paired-samples T tests in a statistical software with the ROC analysis function. RESULTS The fused ensemble model for primaries achieved an area under the curve (AUC) of 0.871 with a sensitivity of 78.3% and specificity of 90.9% at the maximum Youden J statistic, which trended towards improvement when compared to the clinical (AUC = 0.788, P = 0.134) but only reached significance when compared to the radiomic ensemble model (AUC = 0.770, P = 0.017). For nodes, the fused ensemble model achieved an AUC of 0.910 with a sensitivity of 100.0% and specificity of 68.0%, which also trended towards improvement when compared to the clinical model (AUC = 0.865, P = 0.080) but did not reach significance. The most commonly included radiomic features in the ensemble were delta shape features, with change in maximum 3D diameter and change in sphericity being the most common for the primary (69.6%) and nodal (80.8%) models. CONCLUSION The fused ensemble explainable boosting machine model achieved high discriminatory ability at predicting LF of HNSCC in independent primary and nodal structures. Although an additive benefit of delta radiomics over clinical factors could not be proven, the results trended towards improvement with the fused ensemble model, which are promising and worthy of further investigation in a larger cohort.

This paper presents a feature selection method for Internet of Things (IoT) information processing, called MIMIC_FS. The maximal information coefficient (MIC), which can capture a wide range of correlations of variables, including linear, nonlinear, and nonfunctional correlations, is introduced to measure the relevance and redundancy between features and class labels. Based on this measure, the MIC-relevance-average-MIC-redundancy criterion is presented to evaluate the goodness of features, and an approximate-Markov-blanket search strategy is then proposed to improve the efficiency of feature selection. To validate the present study, MIC is also applied to feature selection directly by using the feature-ranking strategy. Experiments on six ASU datasets for IoT applications were conducted. The results show that the proposed method achieves better performance than the comparison methods, markedly reducing feature dimensionality in order to process the tremendous quantities of data in IoT.

Imperceptibility, capacity, robustness and security are four basic requirements of any watermarking technique. A new blind image watermarking technique based on the redundant discrete wavelet transform (RDWT) and singular value decomposition (SVD) is presented in this paper to satisfy all of these four watermarking requirements simultaneously. The gray-scale watermark image is directly embedded into the singular values of RDWT sub-bands after multiplying by a scaling factor. The self-adaptive differential evolution (SADE) algorithm is used to optimize the scaling factor values with the aim of reaching the highest possible robustness while guaranteeing a pre-determined watermarked image quality. By the use of human visual system (HVS) characteristics, an 8-bits digital signature is inserted into the watermarked image to solve the false positive problem which is a prevalent security problem for the most SVD-based watermarking methods. The digital signature is used for verification test before initialization of the watermark extraction procedure. Also, considering the existing redundancy in the RDWT domain, the scheme attained a large amount of capacity. Experimental results demonstrate that in addition to achieving a great imperceptibility, large capacity and sufficient security, the proposed scheme obtains a satisfactory level of robustness against image processing and geometrical attacks, simultaneously.

In this paper, we present a unifying analysis for redundancy systems with cancel-on-start ( c . o . s . ) and cancel-on-complete ( c . o . c . ) with exponentially distributed service requirements. With c . o . s . ( c . o . c . ) all redundant copies are removed as soon as one of the copies starts (completes) service. As a consequence, c . o . s . does not waste any computing resources, as opposed to c . o . c . We show that the c . o . s . model is equivalent to a queueing system with multi-type jobs and servers, which was analyzed in Visschers et al., (2012), and show that c . o . c . (under the assumption of i.i.d. copies) can be analyzed by a generalization of Visschers et al., (2012) where state-dependent departure rates are permitted. This allows us to show that the stationary distribution for both the c . o . c . and c . o . s . models has a product form. We give a detailed first-time analysis for c . o . s and derive a closed form expression for important metrics like mean number of jobs in the system, and probability of waiting. We also note that the c . o . s . model is equivalent to Join-Shortest-Work queue with power of d (JSW( d )). In the latter, an incoming job is dispatched to the server with smallest workload among d randomly chosen ones. Thus, all our results apply mutatis-mutandis to JSW( d ). Comparing the performance of c . o . s . with that of c . o . c . with i.i.d. copies gives the unexpected conclusion (since c . o . s . does not waste any resources) that c . o . s . is worse in terms of mean number of jobs. As part of ancillary results, we illustrate that this is primarily due to the assumption of i.i.d. copies in case of c . o . c . (together with exponentially distributed requirements) and that such assumptions might lead to conclusions that are qualitatively different from that observed in practice.

Feature selection is a preprocessing step in many application areas that are relevant to expert and intelligent systems, such as data mining and machine learning. Feature selection criteria that are based on information theory can be generally sorted into two categories. The criteria in the first group focus on minimizing feature redundancy, whereas those in the second group aim to maximize new classification information. However, both groups of feature evaluation criteria fail to balance the importance of feature redundancy and new classification information. Therefore, we propose a hybrid feature selection method named Minimal Redundancy-Maximal New Classification Information (MR-MNCI) that integrates the two groups of feature selection criteria. Moreover, according to the characteristics of the two groups of selection criteria, we adopt class-dependent feature redundancy and class-independent feature redundancy. To evaluate MR-MNCI, seven competing feature selection methods are compared with our method on 12 real-world data sets. Our method achieves the best classification performance in terms of average classification accuracy and highest classification accuracy.

This paper deals with a static output-feedback stabilization problem within a sampled-data framework for nonlinear systems in Takagi–Sugeno form. We investigate this problem on the basis of the descriptor-redundancy scheme. The resulting features are that: (i) the sampled-data synthesis does not involve (even approximate) discrete-time models of nonlinear systems; (ii) the sampled-data stability is analyzed in the continuous-time Lyapunov sense; and (iii) the design algorithm consists of a single-stage linear matrix inequality problem without linear matrix equality constraints.

Motion compensation is the key technique to reduce temporal redundancy in video coding. Interpolation filters are adopted to generate the inter frame prediction for motion compensation with fractional pixel accuracy. In existing video coding standards such as H.264/AVC and HEVC, a set of predefined interpolation filters is adopted in motion compensation. However, predefined interpolation filters cannot adapt to the video content, which may compromise the coding efficiency. In this paper, a content adaptive interpolation scheme is proposed for motion compensation. In the proposed scheme, a set of adaptive interpolation filters is derived for each frame as additional interpolation filters to minimize the inter prediction difference. Rate-distortion optimization is employed to choose between the predefined interpolation filters and the derived adaptive interpolation filters to achieve the best coding performance at the low bit rates. The proposed scheme is implemented into the HM 12.1 software. Experimental results show that the proposed scheme achieves 5.13 percent, 3.42 percent and 4.07 percent bit rate saving on average compared with HEVC under the “low delay P”, the “low delay B” and the “random access” configurations respectively.

In recent days, Gene Expression Microarray (GEM) data increases its significance to diagnose the diseases appropriately. The key issue of analyzing gene expression information for huge amount of genes with less samples it to extract the disease-related data from the large number of unwanted information and noise. Feature choice, removing unwanted and inappropriate genes was a major process for investigating this challenge. So, the Feature Selection (FS) and categorization are the primary tasks in analyzing the GEM. In this article, a Swallow Swarm Optimization (SSO) with Score-based Criteria Fusion (Optimized SCF (OSCF)) wrapper FStechnique is developed to predict the cancer with a high classification efficiency. Novel wrapper FStechnique follows the procedure of Binary Weight SSO (BWSSO) and the fitness is evaluated by combining the SCF and classification accuracy. OSCF algorithm fuses SCF and classification accuracy to measure the genes significance and redundancy for verifying the genes dependency. It could improve the efficiency of the classifiers such as K-Nearest Neighbours (KNN), Support Vector Machine (SVM) and Recursive Neural Networks (RNN), and evaluate its improvement using four different GEM datasets (Prostate cancer, Small Round Blue Cell Tumours (SRBCT), Leukemia, and Lymphoma). Experimental results verify that the OSCF technique shows better results when compared to conventional techniques based on precision, recall, accuracy and Area Under Curve (AUC).

Accurate and reasonable wind speed prediction system has a significant impact on the utilization of wind energy. A novel combination forecasting model based on Long Short-Term Memory (LSTM) network and BP neural network is designed in this paper. This model combines the principle of deep learning algorithm and the improved BP neural network to deal with nonlinear wind speed prediction. Before the prediction, singular spectrum analysis (ssa) and complete ensemble empirical model decomposition adaptive noise (CEEMDAN) are selected as the data pretreatment part to de-noise the original wind speed data and decompose it into multiple components. This part is conducive to improving the signal-to-noise ratio (SNR) of wind speed data and simplifying the characteristics of wind speed data. Then, in order to reduce the error accumulation and computation redundancy, fuzzy entropy (FE) is used to calculate the time complexity of each component, according to the Spearman correlation, the inherent mode function (IMF) components are recombined to form a new subsequence. Experimental results show that the error accumulation can be reduced by 48.65% for dataset 1 and 29.53% for dataset 2, and the operation time can be reduced by about 50% for two datasets. To avoid the limitation of a single model, introducing the LSTM and improved BPNN which improved by sparrow search algorithm (SSA) two different prediction models are used to predict the sub sequences with high complexity and the low complexity subsequences, respectively. Finally, the predicted values of the models are superimposed to get the final values. In order to verify the validity of the proposed model, the final predictions, compared with six different prediction models, show that this model can achieve the best performance and obtain higher prediction accuracy. Such as the performance evaluation indexes (RMSE = 0.051, MAPE = 0.929%) are smallest obtained from dataset1 by one-step prediction, and (RMSE = 0.086, MAPE = 0.966%) are smallest obtained from dataset2 by one-step prediction. In addition, the Pearson correlation between the predicted value and the true wind speed value obtained by the prediction model applied to the two data sets is the highest 99.17% and 98.73%, respectively.

A random-forest (RF)-assisted artificial-neural-network (ANN) method (RF-ANN method) was developed to improve the analytical performance of laser-induced breakdown spectroscopy (LIBS) in evaluating the composition of low-alloy steel. The RF algorithm was employed to identify the weight values of spectral variables in the quantitative analysis, and the weight values were used as the criteria to select the spectral variables. The selected variables were then used as inputs to train the ANN, which was then used as the analysis model. Compared to the calibration curve method, the RF-ANN method performs well in nonlinear fitting and utilizes an abundance of spectral information; moreover, it can accommodate large amounts of plasma data via filter spectral variables, thus overcoming the problem of the ANN training speed being too slow for practical use due to the redundancy in spectral variables. The results revealed an improvement in the analytical performance of LIBS using the RF-ANN method to predict the concentrations of chromium (Cr) and vanadium (V) in low-alloy steel samples. Specifically, the root-mean-square errors of cross-validation (RMSECV) for Cr and V were reduced from 0.060 and 0.042 wt% using the calibration curve method to 0.021 and 0.007 wt%, respectively, using the RF-ANN method.

Blind image quality assessment (BIQA) aims to automatically evaluate image quality without any prior knowledge of reference images and distortion types. Most of existing BIQA methods use certain probability distribution models to capture the natural scene statistics features of images in bandpass domains which can be viewed as a process of removing redundancy. There also exist methods which apply NSS features in redundancy domain. In this paper, we propose a novel method that employs both bandpass and redundancy domains to acquire the complementary features in multiple color spaces. Furthermore, hierarchical feature extraction strategy is adopted to make the image representation more powerful. Then we stack them as a multi-channel feature maps group, and use Gaussian mixture model to fit them. Finally, Fisher Vectors are used to encode them and a support vector regression model is trained as the quality predictor. Extensive experiments on four commonly evaluated image quality assessment benchmark databases show the proposed method is very competitive against other BIQA methods and has good generalization ability.

While web services avail a rapid growth of data volume for use, identifying helpful information is of great value, especially when users face with an unwilling glut of information. Thus, it is deemed relevant and meaningful to provide users with a representative subset (i.e., small set) that could well reflect the original information corpus (i.e., large set). In such a large–small context, this paper addresses the issues of representativeness in light of measurement and extraction by reviewing our previous efforts. Specifically, we first discuss various metrics from different perspectives of representativeness, then present a series of related representativeness extraction methods. Finally as a supplement and extension, a recent effort is introduced, which aims to take information quality into account in deriving a ranked subset. The proposed extraction method is justified by extensive real-world data experiments, showing its superiority to others in both effectiveness and efficiency.

For band selection in hyperspectral images, there are two crucial aspects, information preservation and redundancy reduction, which have a great influence on the subsequent applications. Most of current works have a fixed preference on these two aspects. However, since different data sets possess different numerical characteristics, optimal preferences on these two aspects may be different and difficult to be decided. Therefore, a method is required, which can explore optimal trade-offs of these two aspects according to different characteristics of data sets. To address this challenge, a novel multi-objective optimization model for band selection is proposed. Two conflicting objective functions compose the proposed model. One measures the amount of information and the other one measures the redundancy contained in the selected bands. Through this model, the two aspects are quantified, which makes it possible to optimize them simultaneously. To optimize this model, a new multi-objective immune algorithm is designed to fit the characteristics of hyperspectral data. And it has the ability to obtain a series of Pareto optimal solutions which represent different optimal trade-offs between the two objective functions. In this way, the proposed method can explore the optimal trade-offs between the two aspects and provide decision makers more options according to different characteristics of data sets. Experiments are implemented on three real hyperspectral data sets. The results show the superiority of the proposed method.

This paper proposes a multilevel reversible data hiding scheme in encrypted domain by utilizing the controllable redundancy of learning with error public key cryptography. Messages can be embedded into multilevel sub-regions of ciphertext by quantifying the encrypted domain and recoding its redundancy. We recode redundancy based on the characteristics of cipher’s distribution. Extraction and decryption processes are separated by dividing the encrypted domain into multilevel sub-regions and introducing different quantification standards. Original plaintext can be losslessly recovered from the marked ciphertext by using the decryption key; with a specific level data-hiding key, only the message hiding in the corresponding level can be extracted, while plaintext and other messages remain secret. We provide theoretical analysis and experimental results on the feasibility, reversibility, and security of the proposed scheme. The capacity and encryption blow up factor are discussed. The experimental results demonstrate the maximum embedding rate can exceed 0.3000 bpb of ciphertext.

Query diversification as an effective way to reduce query ambiguity and enhance result novelty has received much attention in top-k search applications on large networks. A major drawback of the existing diversification models is that they do not consider redundancy elimination during the course of search, resulting in unassured novelty in the search result. In this paper, to improve the novelty of the search result, we propose a new method of diversified top-k similarity search by combining diversification of node attribute coverage with a dissimilarity constraint. Due to the non-monotonicity implied by the dissimilarity constraint, existing techniques based on monotonicity assumptions cannot be applied. Our model requires solving a new problem of Dissimilarity Constrained Non-monotone Submodular Maximization (DCNSM). Based on constructing a dissimilarity-based graph, we solve this problem by a greedy algorithm achieving an approximation ratio of 1/Δ, where Δ is the maximum node degree of the dissimilarity-based graph, in time linear to the number of edges of the graph. We show that DCNSM cannot be approximated in ratio | V | 1 − ϵ , indicating that our solution achieves an optimal ratio. We conduct extensive experiments on both synthetic and real-world attributed network datasets. The results show that our diversification model significantly outperforms the baseline methods, and confirm that combining dissimilarity constraint in diversification can significantly improve the novelty of search result.

In general, in order to reduce the number of bands in hyper-spectral images, transformed features such as principal component analysis (PCA) cannot directly provide information indicating which raw spectral bands are essential, even though they provide reduced dimensionality for the consecutive analysis. This paper proposes minimum redundancy and maximum relevance (mRMR) feature selection techniques to directly choose essential raw bands from hyper-spectral images. In addition a deep neural network is suggested to classify the hyper-spectral data with reduced dimension, which consists of a convolutional neural network (CNN) followed by a fully connected network (FCN). The CNN extracts meaningful features like PCA, but in a nonlinear, supervised manner, and the FCN classifies the six apple leaf conditions including normal, young, malnutrition, early and late stages of apple Marssonina blotch (AMB), and background. Experimentally, we found five essential spectral bands using the mRMR techniques (777.24 nm, 547.77 nm, 474.32 nm, 859.45 nm, and 735.85 nm), which proved to have better accuracy than RGB image for the classification using the deep neural network. The proposed scheme is applicable for band reduction to obtain the efficient multispectral sensor system in the sense that only several essential spectral bands are chosen, and more accurate diagnosis of plant diseases is possible by reducing the complexity involved in hyper-spectral images.

In this paper, a reversible data hiding scheme for vector quantization (VQ)-compressed images is proposed. The scheme introduces two methods to explore the high correlations between VQ indices of an index table, followed by exploiting the redundancy elimination for data embedding. The first method utilizes the concept of side matching and the second one uses the feature of locally repetitive occurrence of the VQ indices in VQ-compressed images. In this research, some concepts of information theory are studied to design an efficient index mapping mechanism in the embedding capacity point of view. The proposed index mapping mechanism partitions the positions of the sorted codebook into some intervals, and assigns each interval of the positions to each position with high hit rate in the sorted codebook. Then, based on the secret bits, each index located in the positions of the sorted codebook with high hit rates is mapped to one of the indices in the interval assigned to that position. Furthermore, an interval of the sorted codebook is reserved as the indicator of the indices with low hit rates. The experimental results demonstrate that the proposed scheme significantly outperforms the existing schemes in terms of the embedding capacity and compression-embedding efficiency.

Pixel value grouping (PVG) has been proved effective in exploiting block redundancy. However, the uniform pixel block still exposes the deficiency in generating prediction errors in a block-by-block manner. In this paper, a novel pixel-based pixel value grouping (PPVG) is proposed with which the embedding primitive switches from pixel block to pixel and PVG is fulfilled on context pixels to calculate predicted value. In this way, not only the advantage of PVG can be preserved but also all pixels can be fully utilized and classified. Then more and smaller expandable prediction errors can be obtained by utilizing pixels in smooth regions with priority. As a result, the embedding procedure can embed more data with less distortion. Experimental results verify that the proposed scheme outperforms other state-of-the-art schemes.

In this paper, we propose an efficient algorithm for dynamic magnetic resonance (MR) image reconstruction. With the total variation (TV) and the nuclear norm (NN) regularization, the TVNNR model can utilize both spatial and temporal redundancy in dynamic MR images. Such prior knowledge can help model dynamic MRI data significantly better than a low-rank or a sparse model alone. However, it is very challenging to efficiently minimize the energy function due to the non-smoothness and non-separability of both TV and NN terms. To address this issue, we propose an efficient algorithm by solving a primal-dual form of the original problem. We theoretically prove that the proposed algorithm achieves a convergence rate of O(1/N) for N iterations. In comparison with state-of-the-art methods, extensive experiments on single-coil and multi-coil dynamic MR data demonstrate the superior performance of the proposed method in terms of both reconstruction accuracy and time complexity.

In difference/prediction error expansion, an efficient capacity-distortion trade-off generally indicates excellence in embedding performance. From this point of view, existing schemes which generate difference/prediction error in block-by-block manner are far from efficient and uniform-blocking may be the main reason. To overcome this drawback, a novel multi-sub-blocking (MSB) strategy which enables three or even more various-sized blocks based embedding is proposed in this paper. After initial division, the dispersion of pixel values within block is predicted to identify exceptional block which may produce to-be-shifted pixels. Then further division would be inevitable for those blocks to obtain closer correlation of pixels within smaller block to achieve improved block redundancy mining. To verify the advantage of MSB, this work extends two existing RDH schemes by incorporating MSB. Experimental results demonstrate that the two proposed schemes both achieve improved embedding performance and outperform previous related state-of-the-art schemes.

In modern communication network such as a public Internet, random losses of information can be recovered by frame expansions. In this paper, we think about the transport protocol that delivers a coefficient of frame expansions with some probability of failure via multiple intermediate nodes. By this protocol, we abstract the transmission as a Markov process. In this case, even if the probabilities of erasures are close to the 1/2, we show that the reconstruction error can be minimized as small as possible by using frames with high redundancy. Furthermore, we set up a probability model for constructing optimal Parseval frames that are robust to one probabilistic erasure. It is shown that probabilistic modeled Parseval frames are more effective than traditional Parseval frames on recovering the signal with probabilistic erasures.

Redundancy ensures reliable data for decision making. Reliable data plays a very important role in the analysis, monitoring and forecasting of system behaviour whereas bad quality data may provide erroneous result in decision scheme. In Wireless Sensor Network (WSNs), nodes are densely deployed in a region to collect information. Sensors sense the similar data and forwards to sink. This similar data sometimes leads to redundancy at the sink. The redundant data results in more accuracy, reliability and security whereas elimination helps in energy saving as most of the energy of sink node gets waste in dealing with the redundant data. Data accuracy still needs to be preserved even if there is increase in network cost and/or time. Therefore, there is requirement of a mechanism in which we can extract information from the redundant data and be able to provide a more consistent, accurate and reliable data set in an energy efficient manner. The data fusion techniques help in maintaining the same. This paper presents the various data fusion approach that shows the impact of redundancy in the area of WSNs.

Feature selection is an essential step in various tasks, where filter feature selection algorithms are increasingly attractive due to their simplicity and fast speed. A common filter is to use mutual information to estimate the relationships between each feature and the class labels (mutual relevancy), and between each pair of features (mutual redundancy). This strategy has gained popularity resulting a variety of criteria based on mutual information. Other well-known strategies are to order each feature based on the nearest neighbor distance as in ReliefF, and based on the between-class variance and the within-class variance as in Fisher Score. However, each strategy comes with its own advantages and disadvantages. This paper proposes a new filter criterion inspired by the concepts of mutual information, ReliefF and Fisher Score. Instead of using mutual redundancy, the proposed criterion tries to choose the highest ranked features determined by ReliefF and Fisher Score while providing the mutual relevance between features and the class labels. Based on the proposed criterion, two new differential evolution (DE) based filter approaches are developed. While the former uses the proposed criterion as a single objective problem in a weighted manner, the latter considers the proposed criterion in a multi-objective design. Moreover, a well known mutual information feature selection approach (MIFS) based on maximum-relevance and minimum-redundancy is also adopted in single-objective and multi-objective DE algorithms for feature selection. The results show that the proposed criterion outperforms MIFS in both single objective and multi-objective DE frameworks. The results also indicate that considering feature selection as a multi-objective problem can generally provide better performance in terms of the feature subset size and the classification accuracy.