Search

In this paper, we propose an abnormal event detection method based on the combination of Riemannian manifold and LSTM network. We divide the video into a series of non overlapping regions, and extract HOG and HOF as original appearance and motion features. Because the existing abnormal event detection methods pay little attention to the changes rate of appearance and motion, we propose a new feature called Riemannian manifold distance feature that can represent the topological relationship of features at different moment in the same position and capture the change rate of the current feature. Once the change rate of the current features changes, the topological relationship of features will change. We use ISOMAP algorithm to embed manifold into low dimensional Euclidean space. In the training phase, the dimension-reduced features of the same position at different historical moment are used as the input of each time step of the LSTM network, and the dimension-reduced feature of the current moment is used as expected output. And in this stage, we only use the normal video, so the LSTM will learn the changing rules of the features contained in normal events. In the testing phase, we use LSTM network to predict current dimension-reduced HOG and HOF features. After that, we can obtain the Riemannian manifold distance feature of actual dimension-reduced feature and predicted dimension-reduced feature. In order to detect abnormal events, we propose an abnormal score, which is used to measure whether the appearance/ movement and the change rate of them are close to the expectation. The abnormal score can detect the abnormal event according to dimension-reduced appearance feature, dimension-reduced motion feature, appearance Riemannian manifold distance feature and motion Riemannian manifold distance feature. Experiments on four benchmark datasets show the competitive performance of our method with the state-of-the-art methods. [ABSTRACT FROM AUTHOR]

• Gaussian Processes strongly depend on kernel functions in any application. • A kernel can be represented as a tree of basic mathematical expressions. • In this work, Genetic Programming is proposed to learn such expression trees. • This approach has been tested in several time-series forecasting problems. Choosing the best kernel is crucial in many Machine Learning applications. Gaussian Processes are a state-of-the-art technique for regression and classification that heavily relies on a kernel function. However, in the Gaussian Processes literature, kernels have usually been either ad hoc designed, selected from a predefined set, or searched for in a space of compositions of kernels which have been defined a priori. In this paper, we propose a Genetic Programming algorithm that represents a kernel function as a tree of elementary mathematical expressions. By means of this representation, a wider set of kernels can be modeled, where potentially better solutions can be found, although new challenges also arise. The proposed algorithm is able to overcome these difficulties and find kernels that accurately model the characteristics of the data. This method has been tested in several real-world time series extrapolation problems, improving the state-of-the-art results while reducing the complexity of the kernels. [ABSTRACT FROM AUTHOR]

Many RGBT trackers utilize adaptive weighting mechanism to treat dual modalities differently and obtain more robust feature representations for tracking. Although these trackers work well under certain conditions, however, they ignore the information interactions in feature learning, which might limit tracking performance. In this paper, we propose a novel cross-modality message passing model to interactively learn robust deep representations of dual modalities for RGBT tracking. Specifically, we extract features of dual modalities by backbone network and take each channel of these features as a node of a graph. Therefore, all channels of dual modalities can explicitly communicate with each other by the graph learning, and the outputted features are thus more diverse and discriminative. Moreover, we introduce the gate mechanism to control the propagation of information flow to achieve more intelligent fusion. The features generated from the interactive cross-modality message passing model will be passed selectively through the gate layer and concatenated with original features as the final representation. We extend the ATOM tracker into its dual-modality version and combine it with our proposed module for final tracking. Extensive experiments on two RGBT benchmark datasets validate the effectiveness and efficiency of our proposed algorithm. [ABSTRACT FROM AUTHOR]

This paper focuses on the following problem: Given a large number of unlabeled face images, group them into individual clusters, and the number of clusters cannot be known in advance. To this end, an Efficient Density-based clustering incorporated with the model of Graph partitioning (EDG) is proposed. 1. Inspired by the progress of graph partitioning clustering, a novel criterion that can be seen as a variant of the Normalized-cut model is employed to measure the similarity between two samples. 2. We only consider the similarities and connections on a subset of all possible pairs, i.e. the top- K nearest neighbors for each sample. Therefore, the computing and storage costs are linear w.r.t. the number of samples. In order to assess the performance of EDG on face images, extensive experiments based on a two-stage framework have been conducted on 19 benchmark datasets (14 middle-scale and 5 large-scale) from the literature. The experimental results have shown the effectiveness and robustness of our model, compared with the state-of-the-art methods. [code] [ABSTRACT FROM AUTHOR]

• An effective compression algorithm was proposed for real-time transmission data. • A mixed prediction model of LSTM and XGBoost was established for time series data. • High performance of the proposed algorithm was veried by comparative experiments. • Almost 50% compression rate can be reached under one in ten thousand distortions. In this paper, a compression and decompression algorithm is proposed for real-time transmission data by combining discrete wavelet decomposition and reconstruction, hybrid model prediction of long short-term memory (LSTM) and extreme gradient boosting (XGBoost), and the development of quantization sequences. The forward compression and reverse decompression are achieved by continuously acquiring the prediction error and data series with quantization error through the iterative shifting approach. To accurately forecast data series with quantization errors, three classes of data with specific distribution characteristics collected from the measurement-while-drilling (MWD) operation field are decomposed into approximate data and detailed data, which are then predicted by autoregressive integrated moving average (ARIMA), support vector regression (SVR), XGBoost, backpropagation neural network (BPNN), radial basis function neural network (RBFNN), and LSTM models. Particularly, the LSTM model and the XGBoost model are proven to be suitable strategies for working with approximate data and detailed data separately by comparing with conventional approaches in terms of the mean absolute error (MAE), root mean square error (RMSE), and coefficient of determination (R 2) predictive indexes. Besides, experiment studies show that almost 50% compression performance can be reached under one in ten thousand distortions. It is verified that the proposed algorithm is appropriate and efficient for real-time data transmission applications with high performance. [ABSTRACT FROM AUTHOR]

This paper focuses on the design of online event-triggered optimal control strategy for multi-player zero-sum games (MP-ZSGs) with control constraints when the system model is partially unknown. Non-quadratic functions are utilized to construct the cost functions under the condition of control constraints. The proposed algorithm is designed based on the framework of identifier-critic networks. The unknown drift dynamics model is reconstructed by an identifier neural network (INN) using the input and output data. The near-optimal event-based controls and time-based disturbances are designed by training a critic neural network (CNN). With the aid of the designed event-triggered mechanism (ETM), the needless computing and communication actions of the system signals have been reduced so as to save computing/communication resources. Meanwhile, to remove the persistence of excitation (PE) condition, the historical and current data are utilized to construct a modified tuning law of CNN. Theoretically, the uniform ultimate boundedness (UUB) properties of the system states and the critic weights errors are proved by Lyapunov approach. Moreover, the Zeno behavior is proved to be excluded under the designed triggering condition. Finally, the convergence and performance of the online method are verified by simulating a representative example. [ABSTRACT FROM AUTHOR]

Clustering is an important research direction in data mining. However, there is no one clustering algorithm that can be applied efficiently in all situation. Clustering ensemble is the best way to solve the above-mentioned problems. It combines the results of multiple clustering algorithms, and the final result is significantly better than a single clustering algorithm. Although there is a lot of constraint information, the existing clustering ensemble algorithm does not utilize it. This paper uses constraint information in consensus function and proposes a Semi-supervised Selective Clustering Ensemble based on Chameleon (SSCEC) and Semi-supervised Selective Clustering Ensemble based on Ncut (SSCEN) to solve the above problem. SSCEC uses the chameleon algorithm as consensus function, and processes constraint information in subgraph partition and subgraph combining. SSCEN uses the Normalized cut algorithm as consensus function, and processes constraint information in the process of graph dichotomy. The experiment results show that our proposed two semi-supervised member selection clustering ensemble algorithms are better than other semi-supervised algorithms. [ABSTRACT FROM AUTHOR]

Normalized least mean squares (NLMS) and affine projection algorithm (APA) are two successful algorithms that improve the stability of least mean squares (LMS) by reducing the necessity to change the learning rate during the training process. In this paper, we extend them to multi-layer neural networks. We first prove that it is possible to consider a multi-layer neural network as a stack of adaptive filters. It opens the door to bring successful algorithms from adaptive filters to neural networks. We additionally introduce a more comprehensible interpretation than the complicated geometric interpretation in APA for a single fully-connected (FC) layer that can easily be generalized, for instance, to convolutional neural networks and mini-batch training. With this new viewpoint, we introduce a more robust algorithm by predicting and then amending the adverse consequences of some actions that take place in mini-batch backpropagation (BP), even before they happen. The proposed method is a modification to the BP that can be used alongside stochastic gradient descent (SGD) and its momentum variants like Adam and Nesterov. Our experiments show the usefulness of the proposed method in the training of deep neural networks. It is less sensitive to hyper-parameters and needs less intervention during the training process. Besides, it usually converges more smoothly and in fewer iterations. Such predictable behavior helps it to get tuned easier, be resilient during the training, and reduce or eliminate its reliance on other techniques such as momentum. [ABSTRACT FROM AUTHOR]

In this paper, an integral reinforcement learning (IRL)-based model-free optimal output-feedback (OPFB) control scheme is developed for linear continuous-time systems with input delay, where the input and past output data are employed rather than the system dynamic model. First, the equivalence between the delayed optimal control and delay-free case is analyzed. Subsequently, the system state is constructed with output signal and the Bellman equation is written in the form of past outputs. Therefore, the IRL algorithm is developed to learn the OPFB control policy, where the iterative policy is evaluated and improved simultaneously. It is proved that the obtained optimal OPFB controller gives the same solution as the optimal state-feedback. Finally, the presented simulation results illustrate the effectiveness of the developed control method. [ABSTRACT FROM AUTHOR]

Lossy image compression is likely to produce blurred images which leads to erroneous image-level understanding. The human visual system (HVS) is focused on the area of interest present in the image. Motivated by this fact, we propose a compression-decompression algorithm focusing on the important content present in different parts of the image. Using the concept of the Guided Grad-CAM (Gradient-weighted Class Activation Mapping) technique to produce heat maps with the help of a heat map generator trained on ResNet-50, a saliency-guided encoding–decoding algorithm is developed. A wider multi-scale saliency guided convolutional neural network (MsSG-CNN) is designed, in which the notion of convolution with different size filters helps to obtain unique but different features. The feature extraction followed by multilevel fusion of features helps deep neural network (DNN) to capture contextual information and obtain high quality, good resolution, visually pleasing images with fine details. The proposed algorithm is tested on the Kodak benchmark dataset, CLIC 2019 challenging dataset, and FDDB facial images dataset. At low bit rates, the MS-SSIM of the proposed algorithm is found to be superior to JPEG, JPEG2000, BPG, WebP, and Minnen's approaches with approximately up to 60%, 24.80%, 11.43%, 23.08% & 75% gains respectively, which is quite a significant improvement, when tested on the Kodak dataset. Similarly, at high bit rates, the improvement in MS-SSIM is approximately up to 41.67%, 37.30%, 23.90% 34.21% & 13.33% when compared with JPEG, JPEG2000, BPG, WebP, and Minnen's approaches respectively. The improvement in PSNR at low and high bit rates is approximately up to 11.32%, 5%, 5.26% and 5.6%, 10.29%, 8.7% as compared to JPEG, Balle's, and Lee's algorithms respectively. The PSNR-HVS has been improved by approximately up to 27.27%, 19.15%, and 28.33%, 28% as compared to JPEG and Toderici's algorithms respectively at low and high bit-rates. A similar type of improvement is obtained with FDDB and CLIC 2019 datasets also, which is discussed in the paper. [ABSTRACT FROM AUTHOR]

• An algorithm that transforms long-tailed problem into hierarchical learning problem. • An effective knowledge transfer paradigm is proposed to help hierarchical learning. • The performance of our model outperforms the state-of-the-art methods. Deep visual recognition methods have achieved excellent performance on artificially constructed image datasets where the data distribution is balanced. However, in real-world scenarios, data distribution is usually extremely imbalanced and exhibit a long-tailed distribution where data in each head class is more than the class in the tail. Many efficient deep learning methods fail to work normally, i.e., they perform well in the head class while poor in the tail class. In this paper, we propose a two-layer Hierarchical-Learning Long-Tailed Recognition (HL-LTR) algorithm which transforms the long-tailed problem into a hierarchical classification problem by constructing a hierarchical superclass tree in which each layer corresponds to a recognition task. In the first layer of the tree, the degree of data imbalance is largely decreased. The recognition task of the second layer is the original long-tailed recognition problem. The training of HL-LTR is top-down. The knowledge learned by the first layer transfers to classes of the second layer and guides the feature learning of the second layer by using attention mechanism module and knowledge distillation method. Compared with directly solving the most difficult long-tailed recognition task, HL-LTR achieves better performance due to its progressive learning method from easy to difficult and effective knowledge transfer strategy. [ABSTRACT FROM AUTHOR]

In vitro microinjection is one of the main approaches to deliver the specimens to the target. However, conventional microinjection is mostly operated by very experienced technicians and the successful rate is relatively low. Moreover, the manual injection process is also very time consuming and tiring because extra attention needs to be paid for precise operation. As a result, the workers suffer from fatigue and the efficiency remains low. To improve the efficiency, researchers have developed automated microinjection systems to free the technicians from this laborious work. Zebrafish has been more popular than ever for the last decade due to its genetic advantages. It is claimed that the gene of Zebrafish highly resembles to human, which results in the fact that more biological and medical practice has been carried out with Zebrafish larvae or embryos. This paper proposes an innovation convolutional neural network based visual processing strategy for biological sample manipulation and microinjection systems. The proposed approach allows high throughput injection with considerable injecting precision. The strategy is to firstly locate a number of biological samples in the visual field, then the proposed visual algorithm identifies the number of the samples and detects the orientation of each sample. After which the samples are rotated one after another by a rotation plate to the desired angle for injection. An optimized rotation strategy is performed to achieve minimum total rotation angle. A path free of collision is also generated by A-star algorithm for the holding pipette. The experiment results show significant improvement in orientation efficiency and the accuracy of multi-sample rotation is enhanced to some extent. [ABSTRACT FROM AUTHOR]

In this paper, the issue of global neural finite-time fault-tolerant control (FTC) is investigated for the medium-scale unmanned autonomous helicopter (UAH). To recognize the actuator bias and loss of effectiveness (LOE) faults, a novel fault detection and identification (FDI) strategy is proposed, which consists of a fault detection observer, two adaptive fault observers and a decision-making algorithm. The neural network (NN) technique is employed to deal with the unknown system uncertainty. In view of the backstepping approach and Lyapunov theory, a finite-time FTC scheme is developed to assure that all closed-loop system tracking errors converge to a small range of zero after a limited amount of time. Meanwhile, by integrating a switching mechanism into the control design, the traditional semi-globally uniformly ultimately bounded (SGUUB) stability is extended to globally uniformly ultimately bounded (GUUB) stability, such that the constraints on initial conditions of the NN controller is moderated. Simulation studies are implemented to demonstrate the usefulness of the presented controller. [ABSTRACT FROM AUTHOR]

K -nearest neighbor (KNN) based weighted multi-class twin support vector machines (KWMTSVM) is a novel multi-class classification method. In this paper, we propose a novel least squares version of KWMTSVM called LS-KWMTSVM by replacing the inequality constraints with equality constraints and minimized the slack variables using squares of 2-norm instead of conventional 1-norm. This simple modification leads to a very fast algorithm with much better results. The modified primal problems in the proposed LS-KWMTSVM solves only two systems of linear equations whereas two quadratic programming problems (QPPs) need to solve in KWMTSVM. The proposed LS-KWMTSVM, same as KWMTSVM, employed the weight matrix in the objective function to exploit the local information of the training samples. To exploit the inter class information, we use weight vectors in the constraints of the proposed LS-KWMTSVM. If any component of vectors is zero then the corresponding constraint is redundant and thus we can avoid it. Elimination of redundant constraints and solving a system of linear equations instead of QPPs makes the proposed LS-KWMTSVM more robust and faster than KWMTSVM. The proposed LS-KWMTSVM, commensurate as the KWMTSVM, all the training data points into a "1-versus-1-versus-rest" structure, and thus our LS-KWMTSVM generate ternary output { - 1 , 0 , + 1 } which helps to deal with imbalance datasets. Numerical experiments on several UCI and KEEL imbalance datasets(with high imbalance ratio) clearly indicate that the proposed LS-KWMTSVM has better classification accuracy compared with other baseline methods but with remarkably less computational time. [ABSTRACT FROM AUTHOR]

Differential privacy, a strict privacy notion, prevents the presence or absence of a single data from being identified. This paper proposes a novel average consensus algorithm to achieve differentially private average consensus on multi-agent systems (MASs) with general directed graphs. For MASs under the proposed algorithm, necessary and sufficient conditions are established on achieving average consensus and preserving differential privacy of the states of the agents at any time instant. Finally, numerical examples are provided to illustrate these results. [ABSTRACT FROM AUTHOR]

Motivated by mixture of probabilistic principal component analysis (PCA), which is time-consuming due to expectation maximization, this paper investigates a novel mixture of probabilistic PCA with clusterings for process monitoring. The significant features are extracted by singular vector decomposition (SVD) or kernel PCA, and k -means is subsequently utilized as a clustering algorithm. Then, parameters of local PCA models are determined under each clustering model. Compared with PCA clustering, SVD based clustering only utilizes the nature basis for the components of the data instead of principal components of the data. Three clustering approaches are adopted and the effectiveness of the proposed approach is demonstrated by a practical coal pulverizing system. [ABSTRACT FROM AUTHOR]

• Transform the max-bisection problem to a linearly constrained optimization problem. • Propose a deterministic annealing neural network algorithm by introducing a barrier function in square-root form. • Prove the existence of the integral minimum solution of the max-bisection problem in theory and the proof is simpler than other algorithms. • Experimental results show that the proposed method outperforms other methods. The max-bisection problem has extensive applications in network engineering, making it imperative to develop effective methods to solve this problem. However, the efficient computation on this problem remains challenging due to its NP-hard computational complexity. To address this question, in this paper, we transform the max-bisection problem to an artificial linearly constrained optimization problem and develop a modified deterministic annealing neural network algorithm by introducing a barrier function. The barrier parameter of this function acts as the temperature in the annealing process and decreases from a large positive number to zero. The initial value of the barrier parameter promises that the artificial problem is convex and can be easily solved at the beginning of the descending process. The solution to the original max-bisection problem is finally obtained through a descent algorithm, in which the bound limit of the solution is always satisfied as long as the iteration step length is less than one. We have proved the theoretical convergence of the proposed algorithm. Numerical results demonstrate that the proposed algorithm has several advantages over both the approximation algorithms and metaheuristic methods. [ABSTRACT FROM AUTHOR]

Multi-view clustering, which utilizes the ample information provided by multiple sources to obtain better performance, has attracted much attention. However, existing clustering algorithms either have no ability to offer confidence for each assignment or suffer from the disturbance of outliers. To address these problems, in this paper, we propose a novel multi-view fuzzy clustering method via transferring softmin to fuzzy models. To obtain fuzzy assignments, we utilize the softmin with temperature and further develop an efficient algorithm to solve the non-convex problem approximately. We also show another explanation for the algorithm from the aspect of the prior distribution of various views. Besides, we design a scalable robust loss function, which interpolates between ℓ 2 -norm and the squared ℓ 2 -norm, to enhance the robustness to outliers. Extensive experiments show the superiority of our model under different clustering metrics. [ABSTRACT FROM AUTHOR]

Today, artificial intelligence is everywhere in people's daily lives. Visual tracking, which is used to identify and continuously track specific targets, is an important research domain in the study of artificial intelligence. However, current visual tracking methods are not accurate enough for object tracking with background clutter, which can easily lead to tracking failures. Therefore, in this paper, in order to solve the problem of tracking failure in clutter background, we propose a template update mechanism to improve the accuracy of visual tracking. First, an original template is saved when the background clutter is detected. During background clutter, we use both the original template and the current template at the location estimated by the optical flow and choose better one. Next, the original template is reused after the background clutter is ended. Finally, the proposed mechanism is used both in the KCF and BACF algorithm to verify the effectiveness of the mechanism. With experiments on the OTB2015 dataset, results show that the proposed mechanism has improved accuracy and success rate of the two baseline algorithms. Meanwhile, in state-of-the-art algorithms, the algorithm using the proposed mechanism also has excellent tracking performance. In addition, this method also has strong tracking robustness and adaptation capability to sequential learning for video data. [ABSTRACT FROM AUTHOR]

Both image decomposition and data completion are not only ubiquitous but also challenging tasks in the study of computer vision. In this paper, different from existing approaches, we propose a novel regularization model for image decomposition and data completion, which integrates relative total variation (RTV) with Schatten- 1 / 2 or Schatten- 2 / 3 norm, respectively. RTV is shown to be able to extract the fundamental structure effectively from the complicated texture patterns and largely to avoid the drawback of oil painting artifacts. Schatten quasi-norm is used to capture texture patterns in a completely-separated manner. The proposed model is in essence divided into "RTV+ double nuclear norm" and "RTV+ Frobenius/nuclear hybrid norm", which can be solved by splitting variables and then by using the alternating direction method of multiplier (ADMM). Convergence of the algorithm is discussed in detail. The proposed approach is applied to several benchmark low-level vision problems: gray-scale image decomposition and reconstruction, text removal, color natural scene image completion, and visual data completion, demonstrating the distinguishable effectiveness of the new model, comparing to the latest developments in literature. [ABSTRACT FROM AUTHOR]

The fitness landscape is a theory applied to the evolutionary dynamics of biological evolution to explain the behavior of evolutionary algorithms in the solution of optimization problems. With the continuous advancement of evolutionary algorithm optimization, a fitness landscape can present more abundant feature information, such as the local fitness, fitness distance correlation, and landscape roughness. These landscape features reflect the optimal solution distribution, quantity, and local unimodal topology of the optimization problem from various angles. This paper expresses the adaptability landscape features of typical optimization problems, engages in a quantitative analysis of the fitness distance correlation information, evaluates the difficulty of solving the problem within the search space, and obtains the correlation degree classification result. The search strategy adapts the mixed mutation and the fitness distance correlation for differential evolution. Empirical studies show that, the fitness distance correlation search strategy for the differential evolution algorithm can avoid falling into the local optimum, improve accuracy and convergence, and solve the single-objective optimization problem in a more comprehensive manner. [ABSTRACT FROM AUTHOR]

• We generalize the usual multiplicative ∊ -indicator I ∊ and additive ∊ -indicators I ∊ + to the forms with L p -norm, i.e. I ∊ p -indicator and I ∊ + p -indicator. • We investigate the properties of I ∊ p -indicator and I ∊ + p -indicator under different p , and find that both I ∊ p -indicator and I ∊ + p -indicator can have a good performance on measuring the population convergence and diversity especially when p is set to be infinity. • We implement a series of comparative experiments to test the performances of I ∊ p -indicator and I ∊ + p -indicator by the proposed I ∊ p -indicator and I ∊ + p -indicator based evolutionary algorithm and compare the proposed indicator-based evolutionary algorithm with five existing algorithms. Experimental results demonstrate that the proposed indicator-based evolutionary algorithm is effective. • Considering the deficiency of the usual IGD metric in measuring the convergence of a population, we propose a more effective performance metric based on I ∊ + p -indicator. In this paper, we generalize the usual multiplicative ∊ -indicator I ∊ and additive ∊ -indicator I ∊ + to the forms with L p -norm, i.e. I ∊ p -indicator and I ∊ + p -indicator, and investigate their properties under different p. From the analysis, we find that both I ∊ p -indicator and I ∊ + p -indicator have a good performance on measuring the population convergence and diversity, especially when p is set to be infinity. In order to show their differences more intuitively, we implement a series of comparative experiments by the proposed I ∊ p -indicator and I ∊ + p -indicator based evolutionary algorithm. The experimental results confirm that all these ∊ -indicators have good performances, particulary as p is set to be infinity. We also compare the proposed indicator-based evolutionary algorithm with five existing algorithms. The experimental results demonstrate that our algorithm is effective. In addition, considering that the usual IGD metric is not accurate in measuring the convergence of a population, we adopt the measure based on I ∊ + p -indicator and propose a more effective performance metric, which is denoted as IGD ∊ + . [ABSTRACT FROM AUTHOR]

This paper studies a smoothing norm objective penalty function for two-cardinality sparse constrained optimization problems. Good properties are proved for the smoothing norm objective penalty functions in solving two-cardinality sparse constrained optimization problems, after which an algorithm is presented. Furthermore, its convergence is also proved under some conditions. The proposed algorithm can solve a satisfactory approximate optimal solution in a numerical example. [ABSTRACT FROM AUTHOR]

This paper proposes a change detection algorithm based on Fully Convolutional Siamese Networks for optical aerial images, which is trained by using an improved Contrastive Loss-Focal Contrastive Loss (FCL). The proposed framework equipped with contrastive loss can extract features directly from image pairs and measure changes by using a distance metric. In other words, this method encourages reducing intra-class variance and enlarging inter-class difference, so that the binarized change map can be obtained by a simple threshold. In change detection task, a critical problem is how to overcome example imbalance (i.e. unchanged examples are much more than changed examples). To address this challenge, a novel focal contrastive loss is proposed to further improve the performance of the model. FCL can reduce the impact of example imbalance and make the model focus learning on hard examples. Extensive experiments demonstrate that the proposed approach is more abstract as well as robust. Compared with other baseline methods, the presented method achieves better results on SAZDA, TISZADOB, CDD, and WHU-CD data set. It achieves state-of-the-art performance in terms of the F1 measure. [ABSTRACT FROM AUTHOR]

Sparse Bayesian learning (SBL) provides state-of-the-art performance in accuracy, sparsity and probabilistic prediction for classification. In SBL, the regularization priors are automatically determined that avoids an exhaustive hyperparameter selection by cross-validation. However, scalability to large problems is a drawback of SBL because of the inversion of a potentially enormous covariance matrix for updating the regularization priors in every iteration. This paper develops an approximate SBL algorithm called ARP-SBL, where the regularization priors are approximated without inverting the covariance matrix. Therefore, our approach can easily scale up to problems with large data size or feature dimension. It alleviates the long training time and high memory complexity in SBL. Based on ARP-SBL, two scalable nonlinear SBL models: scalable relevance vector machine (ARP-RVM) and scalable sparse Bayesian extreme learning machine (ARP-SBELM) are developed for problems of large data size and large feature size respectively. Experiments on a variety of benchmarks have shown that the proposed models are with competitive accuracy compared to existing methods while i) converging faster; ii) requiring thousands of times less memory; iii) without exhaustive regularized hyperparameter selection; and iv) easily scaling up to large data size and high dimensional features. [ABSTRACT FROM AUTHOR]

Lossy image compression is likely to produce blurred images which leads to erroneous image-level understanding. The human visual system (HVS) is focused on the area of interest present in the image. Motivated by this fact, we propose a compression-decompression algorithm focusing on the important content present in different parts of the image. Using the concept of the Guided Grad-CAM (Gradient-weighted Class Activation Mapping) technique to produce heat maps with the help of a heat map generator trained on ResNet-50, a saliency-guided encoding–decoding algorithm is developed. A wider multi-scale saliency guided convolutional neural network (MsSG-CNN) is designed, in which the notion of convolution with different size filters helps to obtain unique but different features. The feature extraction followed by multilevel fusion of features helps deep neural network (DNN) to capture contextual information and obtain high quality, good resolution, visually pleasing images with fine details. The proposed algorithm is tested on the Kodak benchmark dataset, CLIC 2019 challenging dataset, and FDDB facial images dataset. At low bit rates, the MS-SSIM of the proposed algorithm is found to be superior to JPEG, JPEG2000, BPG, WebP, and Minnen's approaches with approximately up to 60%, 24.80%, 11.43%, 23.08% & 75% gains respectively, which is quite a significant improvement, when tested on the Kodak dataset. Similarly, at high bit rates, the improvement in MS-SSIM is approximately up to 41.67%, 37.30%, 23.90% 34.21% & 13.33% when compared with JPEG, JPEG2000, BPG, WebP, and Minnen's approaches respectively. The improvement in PSNR at low and high bit rates is approximately up to 11.32%, 5%, 5.26% and 5.6%, 10.29%, 8.7% as compared to JPEG, Balle's, and Lee's algorithms respectively. The PSNR-HVS has been improved by approximately up to 27.27%, 19.15%, and 28.33%, 28% as compared to JPEG and Toderici's algorithms respectively at low and high bit-rates. A similar type of improvement is obtained with FDDB and CLIC 2019 datasets also, which is discussed in the paper. [ABSTRACT FROM AUTHOR]

In this paper, an integral reinforcement learning (IRL)-based model-free optimal output-feedback (OPFB) control scheme is developed for linear continuous-time systems with input delay, where the input and past output data are employed rather than the system dynamic model. First, the equivalence between the delayed optimal control and delay-free case is analyzed. Subsequently, the system state is constructed with output signal and the Bellman equation is written in the form of past outputs. Therefore, the IRL algorithm is developed to learn the OPFB control policy, where the iterative policy is evaluated and improved simultaneously. It is proved that the obtained optimal OPFB controller gives the same solution as the optimal state-feedback. Finally, the presented simulation results illustrate the effectiveness of the developed control method. [ABSTRACT FROM AUTHOR]

• Developing an under-clustering algorithm to extract the most possible clusters. • Involving heuristics and deep models in the framework for clustering effectiveness. • Making up the deep learning research gap in the field of plant disease annotation. Current annotation for plant disease images depends on manual sorting and handcrafted features by agricultural experts, which is time-consuming and labour-intensive. In this paper, we propose a self-supervised clustering framework for grouping plant disease images based on the vulnerability of Kernel K-means. The main idea is to establish a cross iterative under-clustering algorithm based on Kernel K-means to produce the pseudo-labeled training set and a chaotic cluster to be further classified by a deep learning module. In order to verify the effectiveness of our proposed framework, we conduct extensive experiments on three different plant disease datatsets with five plants and 17 plant diseases. The experimental results show the high superiority of our method to do image-based plant disease classification over balanced and unbalanced datasets by comparing with five state-of-the-art existing works in terms of different metrics. [ABSTRACT FROM AUTHOR]

• ER needs algorithms transfering knowledge of source subjects to a customized model. • User-dependent emotion recognition with PLRSA is a semi-supervised method. • PLRSA is the first to unify the instance reweighting and feature matching paradigm. Recently, many scenarios, such as affective disorders treatment, have sparked rising needs for establishment of personalized emotion recognition (PER) models. Unfortunately, the data sparsity issue violates the basic i.i.d. assumption of supervised learning (i.e., training data and test data are independently and identically distributed). In this paper, we present a semi-supervised joint domain adaption (SSJDA) solution, aiming to inject the hidden domain knowledge from ample labeled data of multiple source individuals into the target subject's customized model. Specifically, we put forward a novel Progressive Low-Rank Subspace Alignment (PLRSA) approach, which unifies a semi-supervised instance-transfer paradigm and an unsupervised mapping-transfer learning paradigm in a single optimization framework. We leverage the boosting-based TrAdaBoost algorithm and the Transfer Component Analysis (TCA) algorithm for the implementation of instance reweighting and feature matching, respectively. Then we introduce the ℓ 2 , 1 - norm to pass feedback and make the joint learning feasible. The central idea is to progressively minimize the cross-domain distribution discrepancies to finally construct the optimal domain-invariant features. We systematically compare the PLRSA method with five state-of-the-art techniques using two public EEG datasets (DEAP and SEED). Both many-to-one and one-to-one evaluations are performed. The experimental results have confirmed the efficacy of the proposed method. [ABSTRACT FROM AUTHOR]

• A HSI restoration method using superpixel segmentation of smooth band is proposed, which exploits the spectral smooth and spatial-spectral similarity of the HSI. • Develop an efficient SSSB algorithm to solve the proposed method. • The method can better remove the mixed noise in the HSI, even for heavy noise. Hyperspectral images (HSIs) are inevitably degraded in the acquisition process by mixed noise including Gaussian noise, impulse noise, stripes, and so on. Recently, many low-rank regularization based HSI restoration methods have been proposed to powerfully remove the mixed noise. However, most of them use the square patch based denoising strategy, which destroyed the boundary information of the objects in the HSI. In this paper, we adopt superpixel segmentation to group the pixels of HSI with adjacent position, similar color, texture and luminance into a homogeneous region, whose shape is adaptive. Several homogeneous regions cover the full HSI. This is better than simply dividing the HSI into square patches. By taking advantage of both the low-rank property and the spectral smooth of the HSI, this approach can efficiently remove the mixed noise with few time. Several experiments verify the performance of the proposed approach for HSI restoration. [ABSTRACT FROM AUTHOR]

Task scheduling has always been one of the crucial problem in cloud computing. With the transition of task types from static batch processing to dynamic stream processing, the dynamic online task scheduling problem has attracted widespread attention. At this stage, explore an effective task scheduling method to implement high quality of service (QoS) requests with limited resources is a considerable challenge. This paper proposes a novel scheduling algorithm called double deep Q-network task scheduling (DDQN-TS), which uses the adaptive learning ability of double deep Q-network (DDQN) to explore the optimal task scheduling strategy. Experiments conducted using the Random, Google, and Alibaba benchmarks to compare several classic algorithms show that the proposed DDQN-TS can guarantee a high task completion rate and efficiently reduce the task average response time. [ABSTRACT FROM AUTHOR]

When solving constrained optimization problems in real industrial processes, both optimality and computational efficiency need to be considered. However, most existing meta-heuristic algorithms are slow to find the global optimum. The first reason is that the way to generate and select candidate solutions is time-consuming. The low probability to generate and select potential solutions in assisting the computational efficiency is another reason. In this paper, a simplified state transition algorithm (STA) and a novel constraint-handling technique are proposed to address the above issues for small size constrained optimization problems. Firstly, three out of four operators in basic STA to produce candidate solutions are selected and two operators are modified with adaptive parameter tuning, which have a large probability to generate potential solutions, but consumes less time. Secondly, the constraint-handling technique considers not only the objective function value and the constraint violation but also the difference among candidate solutions. Thirdly, the sequential quadratic programming embedded into the simplified STA can further speed up the convergence. Experiments are conducted on 22 well-known test functions from IEEE CEC2006 and 4 engineering constrained optimization problems, in comparison with state-of-the-art algorithms. The experimental results show that the proposed method is competitive in finding the optimum faster. [ABSTRACT FROM AUTHOR]

A novel sparse tensor dictionary learning algorithm and a convolutional neural network (CNN) classification method based on this algorithm are proposed for hyperspectral image (HSI) in this paper. The HSI is intuitively represented as a three-dimensional (3-D) cube, and utilizing the joint spatial-spectral information can significantly improve the accuracy of HSI classification. Therefore, the proposed atom-substituted tensor dictionary learning (ASTDL) algorithm utilizes tensor techniques to extract 3-D joint spatial-spectral features from HSI cubes directly. Sparsity constraint is enforced on the coefficient tensors, which obeys the sparsity attribute of HSI. The proposed ASTDL enhanced CNN (ASTDL-CNN) classification method utilizes a two-dimensional (2-D) CNN to extract deep features from the feature tensors obtained by the ASTDL, and to perform the pixel-wise classification. The use of ASTDL, which extracts intrinsic tensor features before CNN, alleviates within-class spatial-spectral variation and reduces the requirement of CNN for the labeled data. Whereas, the CNN works as a 3-D classifier, and provides the final classification results. Besides, we perform the majority vote on the classification map obtained by CNN to refine the classification. The performance of the proposed method is evaluated on three real HSI data sets. The competitive results to the compared state-of-the-art methods demonstrate that the ASTDL-CNN can provide accurate and robust classification results. [ABSTRACT FROM AUTHOR]

• We study an important problem, i.e., recommendation with implicit feedback. • We propose a heterogeneous similarity to capture rich correlations. • We design a novel and flexible recommendation algorithm, i.e., Poi-FHSM. • We conduct extensive empirical studies on five real-world datasets. Recommendation with implicit feedback such as "bought" in e-commerce sites and "like" in online-to-offline services is an important problem because of the abundance of users' online behaviors. Previous works for modeling users' implicit feedback mainly focus on designing some predefined similarity, learned similarity, or hybrid similarity between two users (or two items). However, these similarities either can not capture both the local and global relations among users (or items) simultaneously, or they are not easy to be tuned due to the tradeoff parameter in the linear hybridization. Moreover, information from one single group, either a user group or an item group, may not be sufficient for modeling users' preferences well. To this end, in this paper, we first propose a heterogeneous similarity to capture rich correlations among users and items via a concise integration, and then design a novel recommendation algorithm, i.e., pointwise factored heterogeneous similarity model (Poi-FHSM), in which we fully exploit the proposed similarity in a flexible pointwise preference learning paradigm. Finally, we conduct extensive empirical studies on five real-world datasets, and find that our Poi-FHSM performs better than those based on the predefined, learned and hybrid similarity, as well as the state-of-the-art deep learning-based methods, showcasing the effectiveness of our heterogeneous similarity and recommendation algorithm. [ABSTRACT FROM AUTHOR]

• The first benchmark to evaluate planar object trackers in the natural environment. • Evaluate 22 representative algorithms from four groups. • Analyze the results in detail and provide discussions of future work. Planar object tracking is an important problem in vision-based robotic systems. Several benchmarks have been constructed to evaluate the tracking algorithms. However, these benchmarks are built in constrained laboratory environments and there is a lack of video sequences captured in the wild to investigate the effectiveness of trackers in practical applications. In this paper, we present a carefully designed planar object tracking benchmark containing 280 videos of 40 planar objects sampled in the natural environment. In particular, for each object, we shoot seven videos involving various challenging factors, namely scale change , rotation , perspective distortion , motion blur , occlusion , out-of-view , and unconstrained. In addition, we design a semi-manual approach to annotate the ground truth with high quality. Moreover, 22 representative algorithms are evaluated on the benchmark using two evaluation metrics. Detailed analysis of the evaluation results is also presented to provide guidance on designing algorithms working in real-world scenarios. We expect that the proposed benchmark would benefit future studies on planar object tracking. [ABSTRACT FROM AUTHOR]

This paper investigates the reconstructibility of switched Boolean control networks (SBCNs). Several new types of definition about reconstructibility are proposed where the existence and arbitrariness of the switching signal and the input sequence are inconsistent. A weighted pair graph describing all pairs of states which are indistinguished is defined. A deterministic finite automata method is applied into the reconstructibility analysis. Accordingly, several algorithms are designed to determine these types of reconstructibility. Moreover, for a given reconstructible SBCN, an algorithm for determining the current state is provided. Finally, an example is provided to demonstrate the effectiveness of the proposed algorithms and results. [ABSTRACT FROM AUTHOR]

This paper introduces the idea of learning uniformly ergodic Markov chain for one-against-all support vector machine (OAA-SVM) algorithm. We first obtain the generalization error of OAA-SVM with fast learning rate for uniformly ergodic Markov samples. We also propose a new OAA-SVM method with Markov sampling (OAA-SVM-MS). The experimental researches for benchmark repository confirm that the OAA-SVM-MS algorithm has significantly better performance in sampling and training total time, classification accuracy and the obtained classifier's sparsity compared to the classical OAA-SVM algorithm and other multi-class SVM algorithms. [ABSTRACT FROM AUTHOR]

Semantic segmentation has been a major topic in research and industry in recent years. However, due to the computation complexity of pixel-wise prediction and backpropagation algorithm, semantic segmentation has been demanding in computation resources, resulting in slow training and inference speed and large storage space to store models. Existing schemes that speed up segmentation network change the network structure and come with noticeable accuracy degradation. However, neural network quantization can be used to reduce computation load while maintaining comparable accuracy and original network structure. Semantic segmentation networks are different from traditional deep convolutional neural networks (DCNNs) in many ways, and this topic has not been thoroughly explored in existing works. In this paper, we propose a new quantization framework for training and inference of segmentation networks, where parameters and operations are constrained to 8-bit integer-based values for the first time. Full quantization of the data flow and the removal of square and root operations in batch normalization give our framework the ability to perform inference on fixed-point devices. Our proposed framework is evaluated on mainstream semantic segmentation networks like FCN-VGG16 and DeepLabv3-ResNet50, achieving comparable accuracy against floating-point framework on ADE20K dataset and PASCAL VOC 2012 dataset. [ABSTRACT FROM AUTHOR]

Domain adaptation aims to learn a robust classifier for the target domain by leveraging knowledge from a different source domain. Existing methods realize the alignment of cross-domain distributions in manifold subspace to reduce the distribution divergence between the two domains. However, there exists a conspicuous deficiency in them, i.e., the exploration of preserving statistical and geometrical properties simultaneously is still under insufficient, which, to some extent, would cause the under adaptation effect. The statistical and geometrical properties play an important role in minimizing the domain discrepancy underlying the joint probability distributions. For better and adequately exploiting the statistical and geometrical properties, we propose a novel feature adaptation method in this paper, called domain adaptation with geometrical preservation and distribution alignment (GPDA). Specifically, GPDA performs graph dual regularization in the nonnegative matrix factorization framework with label constraints, to learn the discriminative and domain-invariant features while preserving both the statistical properties and geometrical structures of the original data, such that the cross-domain difference can be effectively and positively narrowed. Meanwhile, GPDA simultaneously aligns the marginal and conditional probability distributions in the nonnegative matrix factorization framework during the learning of domain-invariant features, to further minimize the domain gap between the source and target domains, which can adequately transfer knowledge from the source domain to the target domain. Extensive experiments on seven benchmark datasets demonstrate the effectiveness of the proposed GPDA algorithm in cross-domain image classification. [ABSTRACT FROM AUTHOR]

This paper investigates a new bacterial colony-based feature selection algorithm to improve the classification accuracy of customers for personalized products recommendation. An attribute learning strategy is developed in this study to update the feature related population. Specifically, the features can be weighted according to their historic contributions to both the individual- and group-based subsets. Additionally, the frequency of appearance is also recorded for the feature candidates to improve the diversity of feature distribution and avoid the over-fitting. Based on the weight-based feature indexes and frequency of appearance records, the performance of feature subsets are enhanced by replacing the features being repeatedly appeared in a same vector. To explore the feasibility of the proposed method for the missing feature problems, the objective of the optimization is to minimize the classification error using the acceptable number of features. K-Nearest Neighbor is employed as the learning technique to cooperate with the proposed feature selection method. The effectiveness of the proposed feature selection method is demonstrated by performing test on the datasets from UCI machine learning repository and real-world data from Amazon customer reviews of products. Compared with other seven feature selections methods, the proposed feature selection algorithm outperforms the other algorithms by achieving higher classification accuracy rate using smaller features. [ABSTRACT FROM AUTHOR]

k -Means is a well-studied clustering problem that finds applications in many fields related to unsupervised learning. It is known that k -means clustering is highly sensitive to the isolated points (called outliers). Such outliers can significantly influence the final cluster configuration and should be removed to obtain quality solutions. In this paper, we study the k -means with outliers problem. Given a set of data points, the problem is to remove a set of outliers such that the k -means clustering cost of the remaining points is minimized. Designing efficient algorithms for this problem remains an active area of research due to its important role in dealing with noisy data. We consider a relaxed objective function and propose a local search algorithm for k -means with outliers. It is shown that the algorithm has better performance guarantees than previously implemented methods. In particular, it yields a constant-factor bi-criteria approximation solution to the problem. Moreover, we show experimentally that the algorithm performs much better than its provable guarantee and dominates other state-of-the-art methods for the problem. [ABSTRACT FROM AUTHOR]

Gradient-based algorithms are commonly used for training radial basis function neural network (RBFNN). However, one of the challenges in the training process is determining how to avoid vanishing gradient. To solve this problem, an accelerated gradient algorithm (AGA) is designed to improve the learning performance of RBFNN in this paper. First, an indirect detection mechanism, based on the instantaneous gradient decay rate (IGDR) and instantaneous convergence rate (ICR), is developed to identify the vanishing gradient in learning process. Second, an amplification gradient strategy (AGS), which can increase the gradient value of learning parameters, is designed to accelerate the learning speed of RBFNN. Third, the analysis of AGA-based RBFNN (AGA-RBFNN) is given to guarantee the successful application. Finally, some benchmark and real problems are used to illustrate the effectiveness of AGA-RBFNN. The results demonstrate the effectiveness of AGA-RBFNN. [ABSTRACT FROM AUTHOR]

Facing the challenge of distributed computing on processing large-scale data, this paper proposes a consensus-based decentralized training method with communication compression. First, the decentralized training method is designed based on the decentralized topology to reduce the communication burden on the busiest agent and avoid any agent revealing its locally stored data. The convergence of the decentralized training algorithm is then analyzed, which demonstrates that the decentralized trained model can reach the minimal empirical risk on the whole dataset, without the sharing of data samples. Furthermore, model compression combined with the error-compensated method is considered to reduce communication costs during the decentralized training process. At last, the simulation study shows that the proposed decentralized training with error-compensated communication compression is applicable for both IID and non-IID datasets, and exhibits much better performance than the local training method. Besides, the proposed algorithm with an appropriate compression rate shows comparable performance with decentralized training and centralized training, while saving a lot of communication costs. [ABSTRACT FROM AUTHOR]

In abundant location-based service applications, it is necessary to process continuous spatial keyword queries over geo-textual data streaming. As an important spatial keyword query, the collective spatial keyword (CSK) query aims to find a set of objects such that it covers all the given keywords collectively, the objects within the set are nearest to the query point, and it has the minimum distance between different objects. The existing approaches for the CSK query are mostly index-based algorithms. Although these approaches gain superior performance, their applicability is significantly limited by the necessity to create an index to organize the dataset. Therefore, these index-based approaches cannot be utilized to process data streaming that prevalently exists in most location-based service applications. In addition, the existing algorithms have much room for improvement as the distances between different objects are overlooked when generating feasible candidate sets. Moreover, the results returned by the proposed algorithms could be further refined to offer better decision support for users. In this paper, a greedy algorithm and an approximate algorithm with a provable approximate bound are proposed for the CSK query. Our approaches are appropriate to the CSK queries where the datasets are not suitable to be organized by indexes and can get better query results with less objects and smaller function scores. To boost the query performance, new pruning strategies and heuristic rules are developed. The experimental results demonstrate scalability, efficiency, and effectiveness of the proposed algorithms. [ABSTRACT FROM AUTHOR]

Evolutionary algorithm (EA) incorporating with k-means local search operator represents an important approach for cluster analysis. In the existing EA approach, however, the k-means operators are usually directly employed on the individuals and generally applied with fixed intensity as well as frequency during evolution, which could significantly limit their performance. In this paper, we first introduce a hybrid EA based clustering framework such that the frequency and intensity of k-means operator could be arbitrarily configured during evolution. Then, an adaptive strategy is devised to dynamically set its frequency and intensity according to the feedback of evolution. Further, we develop an opposite search strategy to implement the proposed adaptive k-means operation, thus appropriately exploring the search space. By incorporating the above two strategies, a memetic algorithm with adaptive and opposite k-means operation is finally proposed for data clustering. The performance of the proposed method has been evaluated on a series of data sets and compared with relevant algorithms. Experimental results indicate that our proposed algorithm is generally able to deliver superior performance and outperform related methods. [ABSTRACT FROM AUTHOR]

Asset pricing model is the pillar of modern financial market price theory. It is of great practical and theoretical significance to solve the equilibrium price-dividend function of the asset pricing model. To solve the asset pricing model, this paper develops a novel neural network method called improved trigonometric neural network, which consists of three parts: the improved trigonometric function, the initial-condition extreme learning machine algorithm and the reduction algorithm. First, the equilibrium price-dividend function is described as a stochastic differential equation and this function is translated into a second-order ordinary differential equation equivalently. Second, the improved trigonometric neural network is used to solve the differential equation with initial conditions, where the improved trigonometric function is used to serve as the activation function. The reduction algorithm is proposed to obtain a simpler network structure and a faster computing speed. Third, numerical experiments of the improved trigonometric neural network show that the numerical solution of the price-dividend function will be obtained precisely, quickly and feasibly. Compared with several methods to solve asset pricing model, the improved trigonometric neural network has the highest accuracy and fastest speed. [ABSTRACT FROM AUTHOR]

Abstract Feature (gene) selection of microarray data is a very important and challenging task. This paper proposes a new feature selection method, partial maximum correlation information (PMCI), for microarray data classification. The proposed method extracts several orthogonal components from the feature space to evaluate the importance of each feature. In PMCI method, the component extraction is based on the correlation between feature space and class coding space. Meanwhile, this paper also provides a new generic class-encoding scheme for multi-class problems. This type of encoding scheme can ensure the smooth implementation for PMCI and has a clear geometric meaning. To validate the performance of the proposed method, this paper selects six state-of-the-art algorithms for comparison based on ten widely used microarray benchmark datasets for experiment. The experimental results show that PMCI is an effective feature selection method with good performance and lower time complexity. [ABSTRACT FROM AUTHOR]

Abstract In recent years, convolutional neural networks (CNNs) have shown great performance in various fields such as image classification, pattern recognition, and multi-media compression. Two of the feature properties, local connectivity and weight sharing, can reduce the number of parameters and increase processing speed during training and inference. However, as the dimension of data becomes higher and the CNN architecture becomes more complicated, the end-to-end approach or the combined manner of CNN is computationally intensive, which becomes limitation to CNN's further implementation. Therefore, it is necessary and urgent to implement CNN in a faster way. In this paper, we first summarize the acceleration methods that contribute to but not limited to CNN by reviewing a broad variety of research papers. We propose a taxonomy in terms of three levels, i.e. structure level, algorithm level, and implementation level, for acceleration methods. We also analyze the acceleration methods in terms of CNN architecture compression, algorithm optimization, and hardware-based improvement. At last, we give a discussion on different perspectives of these acceleration and optimization methods within each level. The discussion shows that the methods in each level still have large exploration space. By incorporating such a wide range of disciplines, we expect to provide a comprehensive reference for researchers who are interested in CNN acceleration. [ABSTRACT FROM AUTHOR]

Abstract Dealing with high-dimensional dataset has always been an important problem and feature selection is one of useful tools. In this paper, we develop a new filter based supervised feature selection method by combining Trace Ratio Criterion of Linear Discriminant Analysis (TRC-LDA) and group sparsity regularization. The filter based supervised feature selection method is a classifier-independent method while the TRC-LDA criterion is a recently developed criterion for dimensionality reduction that can well preserve discriminative information of dataset. However, there are seldom methods by utilizing TRC-LDA criterion for feature selection. On the other hand, imposing the l2,0-norm to the projection matrix of TRC-LDA will force some rows in it to be zero while keep other rows nonzero making the index of nonzero rows to be the selected features, however, l 2,0 -nom minimizing problem is NP-hard and intractable. To solve the above problem, in this paper, we develop a new method, namely, Trace Ratio Criterion Discriminative Feature Selection (TRC-DFS), for feature selection. The proposed TRC-DFS has imposed l 2,1 -norm, i.e. an approximation of l 2,0 -norm, to the projection matrix W of TRC-LDA to achieve feature selection. As a result, the proposed TRC-DFS can both achieve feature selection as well as capture the discriminative structure of data. We also extend the proposed method with l 2, p -norm (0 < p < 1) regularization to grasp more sparse property and develop an iterative approach to calculate the optimal solution, which is rigorously proved to be converged. Extensive simulations based on several real-world datasets verify the effectiveness of the proposed methods. [ABSTRACT FROM AUTHOR]

Abstract Twin extreme learning machine (TELM) is an efficient and effective method for pattern classification, based on widely known extreme learning machine (ELM). However, TELM is mainly used to deal with supervised learning problems. In this paper, we extend TELM to handle semi-supervised learning problems and propose a novel Laplacian twin extreme learning machine (LapTELM), which simultaneously trains two related and paired semi-supervised ELMs with two nonparallel separating planes for the final classification. The proposed method exploits the geometry structure property of the unlabeled samples and incorporates it as a manifold regularization term. This allows LapTELM to reap the benefits of fully exploring the plentiful unlabeled samples while retaining the learning ability and efficiency of TELM. Moreover, the paper shows that semi-supervised and supervised TELM can form an unified learning framework. Compared with several mainstream semi-supervised learning methods, the experimental results on the synthetic and several real-world data sets verify the effectiveness and efficiency of LapTELM. [ABSTRACT FROM AUTHOR]