Search

In the paper a new way of pattern semantic interpretation directed at automatic semantic classification and image content understanding will be described. Such an understanding will be based on the linguistic theories of pattern classification and is aimed at facilitation of content analysis for some classes of medical patterns. The approach presented in this paper will show the great possibilities of automatic disease interpretation in some analyzed data. The interpretation will be based on cognitive processes, which imitate the psychological processes of understanding the registered patterns as they take place in the brain of a human being. Cognitive analysis will be based on languages of shape description and picture grammars which allow the creation of syntactic descriptions of selected anatomic organs together with a definition of the semantic meaning of the changes in their shapes. Such descriptions will further allow creating a semantically-oriented representation of visual data describing pattern features which are useful in medical decisions and computer-aided diagnostic systems. This publication presents the cognitive systems of medical data image analysis. Systems that conduct cognitive analyses are designed so that semantic analyses can be used to interpret data and reconstruct medical images. These systems are built for medical types of data analysis systems. An analysis in classical CIAIS (Cognitive Image Analysis Information Systems) will be used to illustrate and present the method of learning new solutions which the systems have no knowledge of. The proposed solution presents the method of building a new class of cognitive systems names E-CIAIS (Extended Cognitive Image Analysis Information Systems).

The self-organizing map has already found appreciation for document classification in the information retrieval community. The map display is a highly effective and intuitive metaphor for orientation in the information space established by a document collection. In this paper we discuss ways for using self-organizing maps for document classification. Furthermore, we argue in favor of paying more attention to the fact that document collections lend themselves naturally to a hierarchical structure defined by the subject matter of the documents. We take advantage of this fact by using a hierarchically organized neural network, built up from a number of independent self-organizing maps in order to enable the true establishment of a document taxonomy. As a highly convenient side effect of using such an architecture, the time needed for training is reduced substantially and the user is provided with an even more intuitive metaphor for visualization. Since the single layers of self-organizing maps represent different aspects of the document collection at different levels of detail, the neural network shows the document collection in a form comparable to an atlas where the user may easily select the most appropriate degree of granularity depending on the actual focus of interest during the exploration of the document collection.

Classification based on image sets has recently attracted increasing interests in computer vision and pattern recognition community. It finds numerous applications in real-life scenarios, such as classification from surveillance videos, multi-view camera networks, and personal albums. Image set based face classification highly depends on the consistency and coverage of the poses and view point variations of a subject in gallery and probe sets. This paper explores a synthetic method to create the unseen face features in the database, thus achieving better performance of image set based face recognition. By considering the high symmetry of human faces, multiple synthetic instances are virtually generated to make up the missing parts, so as to enrich the variety of the database. With respect to the classification framework, we resort to reverse training due to its high efficiency and accuracy. The performance of the proposed approach, Synthetic Examples based Reverse Training (SERT), has been fully evaluated on Honda/UCSD, CMU Mobo and YouTube Celebrities, three benchmark datasets comprising facial image sequences. Extensive comparisons with the other state-of-the-art methods have corroborated the superiority of our approach.

Clustering is used in many applicative fields to summarize information into a small number of groups. Motivated by behavioral extraction issues from urban data, the interest of this paper is to propose a classification method that allows modeling the evolution of cluster profiles over time while considering common regressive effects. The parameters of the proposed model are estimated using variational approximation because maximum likelihood estimation is not suitable in this case. The ability of the model to estimate parameters is evaluated using various simulated data and compared with two other models.

Neural architecture search (NAS) has shown very promising results for automatically designing network models. Most existing cell-based NAS approaches generate the target network model from a source super-network, which usually confront inconsistency issues. In this paper, we propose a new NAS method named STC-NAS, a fast neural architecture search with source-target consistency, so that not only the performance of the searched target model is improved but also the search process is boosted. Specifically, during the search phase, we sample the source super-network to let the samples be consistent with the target model. Moreover, we leverage the Jensen-Shannon divergence to ensure the samples are optimized in the direction of being more similar to the target model. Experimental results demonstrate that our method needs only 0.059 GPU-days to search on CIFAR-10. Benefited from its efficiency, STC-NAS can directly search the target super-network on the target task datasets, achieving 2.42% test error on CIFAR-10, 16.45% test error on CIFAR-100, and 24.2% test error on ImageNet datasets.

Video temporal segmentation plays a vital role in video analysis since many higher-level computer vision tasks rely on it. Some recent efforts have been dedicated to generating temporal action proposals for long and untrimmed videos, which requires methods to generate accurate boundaries for video semantics. In this paper, we propose a novel and efficient Temporal Distribution Network (TDN), to model the long-term distribution of video semantic units (video dictionary). Firstly, we encode the semantics and context relations of video segments with a boundary-specified video embedding method. Then based on temporal convolutional layers, we design a Temporal Distribution Network (TDN) enumerating all the possible temporal locations in one pass and generating proposals that have high action confidence scores by capturing the long-term distributions of video semantics. We validate our method on temporal action proposal generation tasks and action detection tasks. Experimental results on two benchmark datasets, THUMOS14 and ActivityNet-1.3, show that the proposed method can significantly outperform the state-of-the-art approaches. Our model could obtain high-quality action proposals with a much faster speed.

With the rapid development of Internet technology and the explosive growth of digital text, opinion mining has become one of the important research hotspots in the field of natural language processing (NLP). In recent years, neural network based deep learning algorithms have been applied in the field of opinion mining. Considering the relation between temporal and spatial dimensions of text data and the characteristics of natural language itself, traditional deep learning algorithms cannot be comprehensive in the processing of fully feature extraction. In this paper, we propose a new deep learning framework for opinion mining, which includes a temporal feature extraction layer that consists of two layers of bidirectional simple recurrent unit (Bi-SRU) networks extracting features at the word and grammar levels; a semantic feature extraction layer that mainly contains a multi-head attention module; a spatial feature extraction layer with dilated convolution that is used to extract opinion preference features. The Internet movie database (IMDb) is used to verify the performance of the proposed framework. The experiment results show that the proposed framework can effectively improve the classification accuracy, whose performance is better than that of the compared algorithms.

The prediction of dose distribution in volumetric modulated arc therapy (VMAT) plays a curial role in radiotherapy. Accurate VMAT dose prediction for prostate cancer is always a great challenge because of the complexity of the VMAT dose distributions and the low contrast of the organs and tissues in the male pelvic CT images. In this paper, a novel progressive refinement UNet (PRUNet) with rank loss is proposed to address the aforementioned problem. On the one hand, the proposed PRUNet extends the traditional UNet with a novel progressive refinement module to generate more realistic dose distributions. The progressive refinement module generates multiple dose predictions with different resolutions in one forward pass and refines the dose prediction with predicted details at finer levels from lower resolution to higher resolution. On the other hand, it turns out that the dosimetric metrics are sensitive to the order relation among the dose values in dose distributions. A new rank loss function is proposed to optimize the similarity between the order relations among the dose values in dose predictions and that in real dose distributions. The proposed PRUNet is trained in a multi-task framework to jointly learn the dose distribution and the order relation among dose values. To evaluate the performance of the proposed PRUNet, the VMAT dose distributions and anatomical information of 64 prostate cancer patients are collected. The proposed model generates more accurate dose distributions with better quantitative dosimetric metrics than the state of the art UNet models do.

Online transfer learning (OTL) has attracted much attention in recent years. It is designed to handle the transfer learning tasks, where the data of the target domain isn’t available in advance but may arrive in an online manner, which may be a more realistic scenario in practice. However, there typically are two limitations of existing OTL algorithms. 1) Existing OTL algorithms are based on shallow online learning models (SOLMs), e.g., linear or kernel models. Due to this limitation of SOLMs they cannot effectively learn complex nonlinear functions in complicated application and the OTL algorithms based on SOLMs cannot either. 2) Existing algorithms only utilize the latest arrived instance to adjust the model. In this way, the previously arrived instances are not utilized. It may be better to utilize the previously arrived instances as well. In this paper, to overcome the abovementioned two limitations, a buffered online transfer learning (BOTL) algorithm is proposed. In the proposed BOTL algorithm, the learner is designed as a deep learning model, referred to as Online Hedge Neural Network (OHNN). In order to enable the OHNN to be effectively learned in an online manner, we propose a buffered online learning framework that utilizes several previously arrived instances to assist learning. Further, to enhance the performance of the OHNN, a model learned in the source domain is transferred to the target domain. The regret bound of the proposed BOTL algorithm is analyzed theoretically. Experimental results on realistic datasets illustrate that the proposed BOTL algorithm can achieve lower mistake rate than the algorithms compared.

The accurate segmentation of cerebral vessels from time-of-flight magnetic resonance angiography (TOF-MRA) data is crucial for the diagnosis and treatment of cerebrovascular diseases. However, cerebrovascular segmentation remains challenging due to the tiny and complex structure of the vessels. This paper proposed a GAN-based deep-based method that maps the TOF-MRA data to 3D cerebral vessels. To improve the capability of feature expression, the proposed network integrates squeeze-and-excite blocks into a Vnet structure to extract and accumulate features at different scales. In addition, we introduce adversarial loss for network training, which can help the model to learn the distribution of the cerebral vessels. Besides, we introduce island number and minimum diameter to evaluate the spatial consistency and recognized minimal vessel of the cerebrovascular segmentation results. The proposed method is validated on 102 TOF-MRA cases with manually annotated labels. The experimental results indicate that our method can identify more tiny cerebral vessels meanwhile maintaining the spatial consistency of the cerebral vessel structure. Besides, our GAN-based cerebrovascular segmentation method reaches an average dice similar coefficient of 89.89 % , outperforming other state-of-the-art segmentation methods.

This paper proposes an online reinforcement learning(RL) for solving Stackelberg games with completely unknown dynamic systems. To deal with the hierarchical optimal control problem, the key is to find the solution of a two-level optimal control with the leader’s optimal problem being constrained. Firstly, the leader-follower coupled Hamiltonian-Jacobi(HJ) equations subject to the follower’s costate equation is derived. Secondly, an off-policy scheme is designed following the policy iteration(PI) algorithm based on model. The improved off-policy algorithm is given for obtaining the solution to the leader-follower coupled HJ equations. The algorithm is built without requiring any knowledge of the dynamic system, and the leader-follower optimal control policies are directly constructed by the extracted data. Meanwhile, the existence of the Stackelberg equilibrium is demonstrated. Lastly, NNs for each player are built, and NN learning is accomplished online with Kronecker product technique and vectorization, which simples the NN form and could decrease computation burden. Simulation examples are presented to demonstrate the proposed learning algorithm.

Image recognition based on deep learning methods has gained remarkable achievements by feeding with abundant training data. Unfortunately, collecting a tremendous amount of annotated images is time-consuming and expensive, especially in grocery product recognition tasks. It is challenging to recognise grocery products accurately when the deep learning model is trained with insufficient data. This paper proposes multi-angle Generative Adversarial Networks (MAGAN), which can generate realistic training images with different angles for data augmentation. Mutual information is employed in the novel GAN to achieve the learning of angles in an unsupervised manner. This paper aims to create training images containing grocery products from different angles, thus improving grocery product recognition accuracy. We first enlarge the fruit dataset by using MAGAN and the state-of-the-art GAN variants. Then, we compare the top-1 accuracy results from CNN classifiers trained with different data augmentation methods. Finally, our experiments demonstrate that the MAGAN exceeds the existing GANs for grocery product recognition tasks, obtaining a significant increase in the accuracy.

This paper presents a dedicated robust adaptive neural network control (RANNC) scheme for dynamic positioning (DP) of marine vessels with a prescribed performance under model uncertainties, external disturbances and input saturation. To guarantee that the transient performance is always within the prescribed performance constraints, a novel error transfer function is firstly proposed to transfer the original dynamic positioning system with the constrained error behavior into an equivalent unconstrained one. Next, the RANNC scheme is developed with the backstepping technique, and the method of command filter is introduced to avoid the ”dimension disaster” problem. Considering the ship’s inertial matrix and damping matrix are not constant in practice, the adaptive control technique and RBF neural networks are adopted to design the model adaptive controller, which only requires the actuator matrix rather than a completely accurate model. Moreover, minimal learning parameter technique is introduced to minimize the computational burden triggered by weight update of neural networks. Finally, comparison simulation results are provided to illustrate the performance of the proposed RANNC scheme.

In this paper, subject to completely unknown system dynamics and input constraints, a reinforcement learning-based finite-time trajectory tracking control (RLFTC) scheme is innovatively created for an unmanned surface vehicle (USV) by combining actor-critic reinforcement learning (RL) mechanism with finite-time control technique. Unlike previous RL-based tracking which requires infinite-time convergence thereby rather sensitive to complex unknowns, an actor-critic finite-time control structure is created by employing adaptive neural network identifiers to recursively update actor and critic, such that learning-based robustness can be sufficiently enhanced. Moreover, deduced from the Bellman error formulation, the proposed RLFTC is directly optimized in a finite-time manner. Theoretical analysis eventually shows that the proposed RLFTC scheme can ensure semi-global practical finite-time stability (SGPFS) for a closed-loop USV system and tracking errors converge to an arbitrarily small neighborhood of the origin in a finite time, subject to optimal cost. Both mathematical simulation and virtual-reality experiments demonstrate remarkable effectiveness and superiority of the proposed RLFTC scheme.

Human has excellent motor capability and performance in completing various manipulation tasks. During some tasks such as tightening/loosening a screw with a screwdriver, the motion is accompanied by force exertion to the environment (that is, constrained motion). To obtain natural human-robot interaction (HRI) as human interacts/collaborates with the environment, interpreting the human’s intention in a way of motion and interaction force is meaningful for carrying out constrained tasks. This paper proposes a long-short term memory (LSTM) network-based decoding method for the simultaneous estimation of human motion and interactive force from surface electromyography (sEMG) signals. The surface EMG recorded from the muscles of forearm is used to decode human’s motion intention. In order to extract smooth features from non-stationary sEMG signals, Bayesian filter is applied instead of traditional time-domain feature extraction method. From the real-time experiments on eight subjects, the LSTM-based decoding method represents high accuracy of motion estimates (91.7%) and force estimates (96.1%) despite of the existence of muscle coupling and non-stationary mapping during such constrained tasks. It indicates that the estimated motion and interactive force can be further applied for HRI in accomplishing such constrained tasks.

In this paper, the intelligent design for the pursuit-evasion game with large scale multi-pursuer and multi-evader has been investigated. Due to the vast number of agents, the notorious ”Curse of Dimensionality” can seriously challenge the traditional design in multi-player pursuit-evasion game, especially under harsh environment with limited communication resource to support information exchange among multi-players. To address this intractable challenge, the emerging Mean Field Games (MFG) theory has been utilized to solve the optimal pursuit-evasion strategies based on a new form of probability density function (PDF) instead of detailed information from all the other players/agents. As such, not only the information exchange is reduced, but also the computation dimension for the optimal strategy derivation is decreased. Specifically, the MFG has been integrated into the pursuit-evasion game to generate a hierarchical structure where the pursuers and the evaders form two mean field groups separately. To online solve the mean field equations, i.e., two coupled partial differential equations, the actor-critic reinforcement learning mechanism is adopted and further extended to a novel actor-critic-mass-opponent (ACMO) approach. In ACMO, the actor neural network estimates the optimal control, the critic neural network approximates the optimal cost function, the mass neural network learns the agent’s group PDF, and the opponent neural network predicts the opponents’ average states in the form of PDF that causes maximum cost for the agent’s group. The Lyapunov theory is utilized to provide the convergence analysis for all neural networks and the stability analysis for the closed-loop system. Eventually, a series of numerical simulations are conducted to demonstrate the effectiveness of the developed scheme.

In autonomous vehicle technology, environmental perception is a very important part of the whole system. The robustness and accuracy of the perception determine the performance of the whole system. Compared with the camera, lidar can generate point cloud data which can provide accurate 3D information better. At present, more researches focus on environmental perception through point cloud. However, the point cloud has the characteristics of sparsity, unordered and non-uniform distribution, its processing method is also different from the general image-based method. In this paper, we proposed a method of semantic segmentation of point cloud which is realized by convolutional neural network. Through spherical mapping, the scattered point cloud is transformed into 2D spherical map with dense and uniform distribution, which is convenient for the processing of the network. To reduce the loss of point information, the spatial position, reflection intensity information and angle value of point are preserved during spherical mapping. Our work includes a lightweight convolutional neural network that fuses point cloud and image information. In addition, a spatial module is added to supplement the loss of spatial information and improve the performance of segmentation. Finally, a series of comparative experiments and performance evaluation are carried out, which show that the proposed method is not only much better than the original structure, but also better than another image and lidar fusion network in the segmentation of each category.

This paper proposes a novel reinforcement learning-based attitude tracking control strategy for combined spacecraft takeover maneuvers with completely unknown dynamics. One major issue in the context of combined spacecraft attitude takeover control is that the accurate dynamic model is highly nonlinear, complex and costly to identify online, which makes it impractical for control design. To address this issue, we take the advantage of the Q-learning algorithm to acquire the control strategy directly from system input/output measurement data in a model-free manner, and thus the online inertia parameter identification procedure is avoided. More specifically, first, the attitude tracking is formulated as a regulation problem by introducing an argumented system, where the system dynamic model is still required in control design. Then, in order to achieve a model-free control strategy, an online policy-iteration (PI) Q-learning procedure is derived to solve the Bellman optimality equation by utilizing the generated measurement data. In theoretical analysis, it is proved that the iteration sequences of Q value function and control strategy can converge to the optimal ones. In addition, rigorous proof of the stability and monotonicity guarantees of the proposed control strategy are also provided. Furthermore, for the purpose of online implementation, off-policy learning scheme is employed to find the optimal Q value function approximator with neural network structure after data-collection phase. Numerical simulations are exhibited to validate the effectiveness of the proposed strategy.

Guided policy search algorithms have been proven to work with incredible accuracy not only for controlling complicated dynamical systems, but also in learning optimal policies from exploration of various unseen instances. This paper deals with a trajectory optimization problem for an unknown dynamical system subject to measurement noise using expectation maximization and extends it to learning (optimal) policies which have less stochasticity in trajectories because of the higher exploitation efficiency. Theoretical and empirical evidence of learned optimal policies for the new approach is depicted in comparison to some well known baselines which are evaluated on an autonomous system with widely used performance metrics.

Multiple Autonomous Underwater Vehicles (AUV) can complete complex ocean exploration missions cooperatively. This paper proposes a containment control algorithm under time-varying constraints by using a neural network for control of multiple benthic AUVs which are called the Ocean Bottom Flying Node (OBFN) systems. The multiple OBFNs in the presence of nonlinear model uncertainties are under the directed topology. First, we define the auxiliary variable and low-order filter. The anti-windup saturation auxiliary system is constructed in presence of the input saturation. Further, the adaptive law and neural network are designed to compensate environmental disturbances and systems model uncertainties, respectively. Moreover, in order to ensure the control performance of OBFN, an exponential boundary constraint is imposed which could constrain the system error convergent rates and bounds. Lyapunov stability theorem and graph theory are used to prove that the multiple OBFN systems are uniformly ultimately bounded. Finally, simulation results for multiple OBFNs illustrate the effectiveness of the proposed algorithm.

The linear active disturbance rejection control (LADRC) has been applied in many control practices and achieved satisfactory results. However, the controller with fixed parameters cannot achieve the optimal control performance for the controlled plant. In order to optimize the control effort of LADRC with online parameter adjustment, a deep reinforcement learning algorithm, the deep deterministic policy gradient (DDPG), is applied to enhance the LADRC in this paper for the ship course control and to obtain the optimized parameters of LADRC in different conditions. Specifically, the proposed strategy adopts the deep neural network to adjust the control parameters according to the measured states. This makes the proposed method applicable to the nonlinear control systems with high dimensional continuous states and actions. In addition, the stability of the closed-loop system is analyzed based on the Lyapunov method. In simulations, comparisons with Q-learning based LADRC and conventional LADRC controllers are also presented. Simulation results are provided to demonstrate the effectiveness of the proposed method.

In this paper, without dividing the fractional-order complex-valued coupled neural networks (FCVCNNs) into two real-valued systems, the problem of synchronization is investigated for FCVCNNs with time-varying coupling strength. To achieve synchronization, by updating coupling strength, two feasible adaptive protocols are designed, 1) a fractional-order adaptive strategy depending on global information; 2) a fractional-order adaptive law relying on the local information based on the connected dominating set theory. Additionally, instead of making the weighted average technique or the solution of the isolated node equation as synchronization reference value, direct error method is adopted to realize synchronization, which offers a flexible way in analysis of FCVCNNs. At last, a numerical simulation is provided to illustrate the validity of theoretical results.

Person re-identification (re-ID) aims at matching a person-of-interest across various non-overlap cameras with distinguished visual appearance variances. Pre-existing research methods mainly employ deep neural models to train large-scale person re-ID datasets, achieving good performance. However, these methods are primarily deployed only on visual data, which can be easily influenced by the environment variances (e.g., viewpoints, poses, and illuminations). In this paper, we propose an adaptive multi-task learning (MTL) scheme for cross domain and modal person re-ID. It can effectively utilize the visual and language information from multiple datasets for improving learning performance. Comprehensive experiments are also conducted on the widely-used person re-ID datasets, i.e., Market-1501 and DukeMTMC-reID, validating the effectiveness of the proposed method. It can model the domain difference and the relationship between the vision and language modalities and achieve state-of-the-art performance. The source code of our proposed method will be available at (https://github.com/emdata-ailab/Multitask_Learning_ReID).

A sliding mode based fault-tolerant control method using neural learning is studied for hypersonic reentry vehicle (HRV) in this paper. Based on the non-singular second-order terminal sliding mode, the composite neural learning is adopted to deal with the system uncertainties caused by the additive faults. The main point is to construct the prediction error to evaluate the performance of intelligent approximation. Meanwhile, a disturbance observer is employed to estimate the unknown disturbance while the actuator multiplicative fault is compensated with an adaptive law. The simulation of HRV is conducted to verify the effectiveness of the proposed fault-tolerant control scheme.

Despite many decades of research on high-performance Deep Neural Network (DNN) accelerators, their massive computational demand still requires resource-efficient, optimized and parallel architecture for computational acceleration. Contemporary hardware implementations of DNNs face the burden of excess area requirement due to resource-intensive elements such as multipliers and non-linear Activation Functions (AFs). This paper proposes DNN with reused hardware-costly AF by multiplexing data using shift-register. The on-chip quantized log 2 based memory addressing with an optimized technique is used to access input features, weights, and biases. This way the external memory bandwidth requirement is reduced and dynamically adjusted for DNNs. Further, high-throughput and resource-efficient memory elements for sigmoid activation function are extracted using the Taylor series and its order expansion have been tuned for better test accuracy. The performance is validated and compared with previous work for the MNIST dataset. Besides, the digital design of AF is synthesized at 45 nm technology node and physical parameters are compared with previous work. The proposed hardware reused architecture is verified for neural network 16:16:10:4 using 8-bit dynamic fixed-point arithmetic and implemented on Xilinx Zynq xc7z010clg400 SoC using 100 MHz clock. The implemented architecture uses 25% less hardware resources and consumes 12% less power without performance loss, compared to other state-of-the-art implementations, as lower hardware resources and power consumption are especially important for increasingly important edge computing solutions.

In this work, we are interested in a class of Bi-directional associative memory neural networks (BAMNNs) with hybrid time-varying delays and D operator. We establish some sufficient conditions to guarantee the existence and global exponential stability of measure-pseudo almost periodic solutions of the considered BAMNNs, using contraction mapping principle, exponential dichotomy theory, differential inequality techniques, and the properties of the measure-pseudo almost periodic functions. A numerical example along with a graphical illustration are presented to support the effectiveness and fasibility of the obtained results. The results in this paper are original and complement the previous outcomes.

Convolutional Neural Networks (CNNs) have won tremendous success in various applications, such as image recognition and natural language processing. Due to the computing-intensive and memory-intensive features of CNN models, it is challenging to deploy them on devices with limited resources and tight power budgets. To address this limitation, pruning, quantization, weights sharing, and other methods have been proposed as efficient solutions to compress CNN models. However, these works usually do not optimize the hardware design when compressing the CNNs. In this paper, we introduce Conv-Inheritance, a hardware-efficient compression method for CNNs by reducing the number of convolution operations to simultaneously reduce the inference time and hardware computing resources on chip of CNN edge devices. We also develop an extended version, called adaptive Conv-Inheritance, to further improve the inference performance by considering the similarity between pixels. We applied our Conv-Inheritance method to compressing various CNN models and performed extensive experiments on different image datasets. Experimental results demonstrate that Conv-Inheritance is both hardware-efficient and time-efficient on edge devices. Besides, Conv-Inheritance is a plug-and-play compression method for CNNs, and we can run Conv-Inheritance on top of other compression methods without any conflicts.

The process of aligning one image, in the coordinate system of another is called registration. Image registration is vital in the medical domain and is used for diagnosis, therapy planning, and treatment of diseases. Due to its huge applicability and importance, medical image registration has emerged to be a separate domain of research. Beginning from the invasive landmark based registration, innumerable algorithms have been proposed to register two images of the human physiology. However, a breakthrough in the literature occurred when an information theory based measure, mutual information was used to register two images. Since then a large number of new algorithms have developed, which use mutual information for fully automatic registration of medical images. This paper is a survey of these algorithms. Beginning from its development, it discusses about some of the major works done on mutual information based image registration. Some comparative studies with other algorithms have also been discussed. The paper ends with a discussion of the developments in medical image registration algorithms post mutual information, which primarily include deep neural network (DNN) based algorithms.

Edge computing is an emerging computing paradigm, which offers a great opportunity to implement data mining-based services and applications for a large number of devices and sensors in Internet of Things. However, the new paradigm is faced with security and privacy challenges due to the diversity and the limited capability of edge components. In particular, data privacy is one of the most concerned problems for all the participants. In this paper, we propose a framework of privacy-preserving data mining based on private random decision trees in edge computing, which not only gives the strong privacy guarantee, but also provides a certain amount of data utility. Firstly, we design a preservation framework to implement private random decision trees satisfying local differential privacy. Secondly, we present the concrete implementations of algorithms and the corresponding task that each participant needs to undertake. Thirdly, we analyze the key factors to influence privacy and utility, including the allocation of data and privacy budget. Fourthly, we give the improved algorithms to further increase the utility with strong privacy preservation. Finally, extensive experiments demonstrate the good performance of our designed framework.

Neural architecture search (NAS) has made significant progress in recent years. However, existing methods usually search in a small-scale, well-designed architecture space, discover only one architecture in a single search, and hardly rework, severely limiting their potential. In this paper, we propose a novel neural architecture generator (NAG) that can efficiently sample architectures in a large-scale architecture space. Like a generative adversarial network (GAN), our model consists of two components: (1) a generator that can generate directed acyclic graphs (DAGs) as cells or blocks of neural architectures and (2) a discriminator that can estimate the probability that a DAG comes from cells of real architectures rather than the generator. Furthermore, we employ a random search with NAG (RS-NAG) to discover the optimal architecture with our customized requirements. Experimental results show that the NAG can generate diverse architectures with our customized requirements multiple times after one adversary training. Furthermore, compared with the existing NAS methods, our RS-NAG achieves the competitive results with 2.50% and 25.5% top-1 accuracy on two benchmark datasets – CIFAR-10 and ImageNet.

Current traffic congestion and the resulting carbon emissions are two of the main problems threatening the sustainability of modern cities. The challenges facing today's cities focus primarily on the optimization of traffic ow and the transition to electric vehicles. The latter aspect implies the need for an adequate deployment of the infrastructure of charging stations. The inherent complexity in today's cities and the difficulty in implementing new policies whose benefits are difficult to measure and predict has led in recent years to consider the enormous potential of simulation tools and in particular of the agent-based simulation (ABS). ABS allows the speciffication of complex models that reect the complexity and dynamism of urban mobility. Current technology in ABS has evolved and matured sufficiently to provide very sophisticated tools but lacking facilities for a exible and realistic generation of input data in the execution of the experiments. In line with this, this paper introduces two configurable generators that automatize the creation of experiments in agent-based simulations. The generators have been developed with the SimFleet simulation tool enhancing the simulation of realistic movements and location of vehicles, passengers and other users of the urban traffic system within a city. The generators proved to be useful for comparing different distributions of locations as well as different agent movement behaviors based on real city data.

Neural architecture search (NAS) is to automatically design task-specific neural architectures, whose performance has already surpassed those of many manually designed neural networks. Existing NAS techniques focus on searching for the neural architecture and training the optimal network weights from the scratch. Nevertheless, it could be essential to study how to tune a given neural architecture instead of producing a completely new neural architecture in some scenarios, which may lead to a more optimal solution by combining human experience and the advantages of the machine’s automatic searching. This paper proposes to learn to tune the architectures at hand to achieve better performance. The proposed Neural Architecture Tuning (NAT) algorithm trains a deep Q-network to tune neural architectures given a random architecture so that we can achieve better performance on a reduced space. We then apply adversarial autoencoder to make the learned policy be generalized to a different searching space in real-world applications. The proposed algorithm is evaluated on the NAS-Bench-101 dataset. The results indicate that our NAT framework can achieve state-of-the-art performance on the NAS-Bench-101 benchmark, and the learned policy can be adapted to a different search space while maintaining the performance.

This paper investigates noise/fault tolerant incremental algorithms for the extreme learning machine (ELM) concept. Existing incremental ELM algorithms can be classified into two approaches: non-recomputation and recomputation. This paper first formulates a noise/fault aware objective function for nonlinear regression problems. Instead of developing noise/fault aware algorithms for the two computational approaches in a one-by-one manner, this paper uses two representative incremental algorithms, namely incremental ELM (I-ELM) and error minimized ELM (EM-ELM), to develop two noise/fault aware incremental algorithms. The proposed algorithms are called generalized I-ELM (GI-ELM) and generalized EM-ELM (GEM-ELM). The GI-ELM adds k hidden nodes into the existing network at each incremental step without recomputing the existing weights. To have a fair comparison, we consider a modified version of I-ELM as a comparison algorithm. The simulation demonstrates that the noise/fault tolerance of the proposed GI-ELM is better than that of the modified I-ELM. In the GEM-ELM, k hidden nodes are added into the existing network at each incremental step. Meanwhile, all output weights are recomputed based on a recursive formula. We also consider a modified version of EM-ELM as a comparison algorithm. The simulation demonstrates that the noise/fault tolerance of the proposed GEM-ELM is better than that of the modified EM-ELM. Moreover, we demonstrate that the multiple set concept can further enhance the performance of the two proposed algorithms. Following our research results, one can make some non-noise/fault tolerant incremental algorithms to be noise/fault tolerant.

With the rapid development of cloud computing, Internet of things and artificial intelligence, human–computer interaction (HCI) is playing an increasingly important role in the daily life. As an important component of HCI, hand gesture recognition (HGR) system is usually combined with edge computing server, utilizing machine learning, including neural network, decision tree, integrated learning, to achieve low latency and high reliability service. High precision HGR with low computational complexity is prerequisite for the commercialization of gesture recognition. Therefore, this paper proposed a high-precision parameter correction algorithm based on the established scattered-point model and the outlier detection scheme, and a recognition algorithm with multivariable decision tree is then presented for the dynamic hand gestures. The experimental results show that the proposed algorithms can improve the recognition accuracy and effectively reduce the running time, which is conducive to algorithm transplantation and model deployment in edge servers.

Many industrial applications result in the constrained underdetermined nonlinear system (UNS) that needs to be solved. The existing solutions are reported to the constrained UNS with time-invariant coefficients, and they may not perform well for the time-variant case. In this paper, we attempt to address the above limitation by providing a new discrete-time zeroing neural network (DTZNN) for solving the time-variant UNS (TVUNS) under bound constraint. Specifically, the bound-constrained TVUNS (BC-TVUNS) is first transformed into a mixed nonlinear system by introducing a nonnegative variable. Then, a continuous-time ZNN (CTZNN) model is established to solve such a mixed system as well as the BC-TVUNS. By employing a special difference formula to discretize the CTZNN model, the new DTZNN model is thus proposed to obtain the solution of the BC-TVUNS. Theoretical analysis and comparative numerical results are presented to validate the effectiveness of the proposed DTZNN model over the previous DTZNN models. The DTZNN application potential is further indicated on the basis of the simulations on a dual-arm redundant robot using the proposed model.

Distributed generators have been widely used in modern power systems, and microgrids are becoming a key concept for the integration of multiple distributed generators and clusters of loads with great efficiency and high system reliability. In a microgrid, the control strategy is one of the most important issues. Various control schemes have been proposed to achieve the cooperation of distributed generators in a microgrid. This paper surveys cooperative control strategies for multiple distributed generators in DC microgrids. The three popular coordinated control strategies are presented: decentralized control, centralized control, and distributed control. The advantages and drawbacks of the control strategies are compared. Some intelligent control methods such as fuzzy logic control and neural network approaches are also briefly introduced in the context of microgrids. Further research trends and challenges in this area are also discussed.

Occluded person re-identification is a challenging task since the available data often suffers from information incompleteness and spatial misalignment. Most state-of-the-art occluded models rely on the external model to provide additional semantic information. However, for the time being, external models, such as the human parsing model and the pose estimation model cannot provide accurate semantic information under a complex occlusion environment and may introduce errors to the Re-ID model instead. In this paper, we propose an occluded person Re-ID model that mines the latent recognizable information of the person image itself, without the help of external models. Feature/Data uncertainty can reduce the influence of noisy samples in datasets and has been discussed in person Re-ID and face recognition, we extend the uncertainty to the micro feature level, and propose the spatial-wise and channel-wise feature uncertainty to constantly refine the features in the spatial domain and the channel domain respectively during feature construction by weakening the influence of noise features. Extensive experiments on the occluded datasets and holistic datasets have proved the effectiveness of our proposed methods.

Machine Learning has emerged in the last years as the main solution to many of nowadays’ data-based decision problems. However, while new and more powerful algorithms and the increasing availability of computational resources contributed to a widespread use of Machine Learning, significant challenges still remain. Two of the most significant nowadays are the need to explain a model’s predictions, and the significant costs of training and re-training models, especially with large datasets or in streaming scenarios. In this paper we address both issues by proposing an approach we deem predictive and user-centric. It is predictive in the sense that it estimates the benefit of re-training a model with new data, and it is user-centric in the sense that it implements an explainable interface that produces interpretable explanations that accompany predictions. The former allows to reduce necessary resources (e.g. time, costs) spent on retraining models when no improvements are expected, while the latter allows for human users to have additional information to support decision-making. We validate the proposed approach with a group of public datasets and present a real application scenario.

This paper shows a computational model that classifies Cybergrooming attacks in the context of COP (child online protection) using Natural Language Processing (NLP) and Convolutional Neural Networks (CNN). The model predicts a high number of false positives, therefore low precision and F-score, but a high accuracy. In this issue, where the number of messages in the context of grooming are so low compared to the number of conversations and messages from other contexts, it can be concluded that is a very consistent and useful result as it captures a high number of true positives, considering that the classifier works for messages. Performing the training of machine learning algorithms with neural networks, semantic analysis and NLP, allows approximate representation of knowledge contributing to discovery of pseudo-intelligent information in these environments and reducing human intervention for characterization of underlying abnormal behavior and detecting messages that potentially represent these attacks.

This paper proposes a sensor network information fusion based distributed state estimation algorithm for tracking non-cooperative maneuvering target. In order to lower the communication burden and energy consumption, an event-triggered mechanism is introduced. The distributed state estimation algorithm for maneuvering target tracking is designed by two stages, namely, local filtering stage and consensus fusion stage. The algorithm proposed in this paper can be used to obtain the high accurate state estimation of the target by introducing a multiple suboptimal fading factor even when the target makes big maneuvering. Moreover, a contribution factor is designed in the information fusion stage to improve the accuracy of maneuvering target state estimation and reduce the consensus iteration times. Besides, the stochastic boundedness of the event-triggered distributed estimation algorithm is proved by introducing a stochastic process. Finally, Monte Carlo numerical simulation example is designed to illustrate the effectiveness of the algorithm.

In this paper, we propose the co-attention dictionary network (CODNet) for weakly-supervised semantic segmentation using only image-level class labels. The CODNet model exploits extra semantic information by jointly leveraging a pair of samples with common semantics through co-attention rather than processing them independently. The inter-sample similarities of spatially distributed deep features are computed to merge reference features through non-local connections. To discover similar patterns regardless of appearance variations, we propose to extract image representations by equipping the neural networks with dictionary learning which provides the universal basis elements for different images. Based on the CODNet model, we propose a multi-reference class activation map (MR-CAM) algorithm which generates semantic segmentation masks for a target image by jointly merging semantic cues from multiple reference images. Experimental results on the PASCAL VOC 2012 and MSCOCO benchmark datasets for weakly-supervised semantic segmentation show that the proposed algorithm performs favorably against the state-of-the-art methods.

Big data will cause the system for business public opinion to be overburdened, and dynamic changes in computing resources will cause tasks to be delayed. To end the situation, this study assigns non-fixed execution time to each task node based on the task soft deadline and task constraints and solves the problem of difficult task scheduling caused by task dependency constraints. Aiming at the problem of task delay caused by dynamic changes of computing resources, this paper proposes an adaptive offloading and scheduling algorithm for dependent tasks in the mobile edge computing environment. This study takes the economic efficiency analysis system as an example and uses the resource allocation management algorithm based on linked lists and edge servers to study the communication resource allocation management of the economic efficiency analysis system. In addition, this study designs experiments to perform performance analysis of the algorithm proposed by this study. The research results show that the proposed algorithm has an obvious effect.

As one of the most complex social problems around the world, crime may bring the risk of dying or losing property to the public if not handled properly. Crime prediction which aims at predicting crime incidents before they happen is of great importance to fight against crime. Previous studies are concerned primarily with day-level crime prediction and have certain limitations on modeling complex spatial–temporal-categorical dependency contained in the criminal activities as well as utilizing external factors to facilitate the forecast. In this paper, we develop a novel Neural Attentive framework for Hour-level Crime prediction (NAHC) to cope with these challenges. Specifically, we first adopt the priori knowledge-based data enhancement strategy to alleviate the zero-inflated issue raised in hour-level settings. Then, multi-graph convolutional networks are applied to capture spatial dependency from different aspects. After that, we integrate gated recurrent units with a temporal attention mechanism to jointly address temporal dependency and capture time-sensitive external factors. A categorical attention mechanism is proposed for dealing with categorical dependency and finally a fully connected network is utilized to generate the final prediction results. Extensive experiments on two real-world crime datasets demonstrate the effectiveness of our framework over the state-of-the-art comparing methods.

A continuous-time distributed algorithm based on multi-agent systems is proposed to solve the convex nonsmooth optimization with globally coupled constraints in this paper. Firstly, the relationship of equivalence between the equilibrium and the optimal solution of the system is proved by using the properties of projection operator and saddle point. In addition, the stability of the algorithm is analyzed by means of Lie derivative and set-valued LaSalle’s invariance principle. In particular, the trajectory of the proposed algorithm from any initial point can converge to the global optimal solution. Finally, the effectiveness of the designed algorithm is verified by numerical simulations.

Edge computing and artificial intelligence (AI), especially deep learning for nowadays, are gradually intersecting to build a novel system, called edge intelligence. However, the development of edge intelligence systems encounters some challenges, and one of these challenges is the \textit{computational gap} between computation-intensive deep learning algorithms and less-capable edge systems. Due to the computational gap, many edge intelligence systems cannot meet the expected performance requirements. To bridge the gap, a plethora of deep learning techniques and optimization methods are proposed in the past years: light-weight deep learning models, network compression, and efficient neural architecture search. Although some reviews or surveys have partially covered this large body of literature, we lack a systematic and comprehensive review to discuss all aspects of these deep learning techniques which are critical for edge intelligence implementation. As various and diverse methods which are applicable to edge systems are proposed intensively, a holistic review would enable edge computing engineers and community to know the state-of-the-art deep learning techniques which are instrumental for edge intelligence and to facilitate the development of edge intelligence systems. This paper surveys the representative and latest deep learning techniques that are useful for edge intelligence systems, including hand-crafted models, model compression, hardware-aware neural architecture search and adaptive deep learning models. Finally, based on observations and simple experiments we conducted, we discuss some future directions. Submitted/Accepted version

Recently, face hallucination, also termed face super-resolution (SR), has been widely studied and achieved significant progress. The algorithm based on manifold learning is one of the primary methods for SR. However, when recovering the high-resolution (HR) counterparts from extremely low-resolution (LR) images, the basic assumption of manifold consistence between LR/HR spaces is doubtful and vulnerable. To address this issue, some algorithms usually employ the cascaded models in line with the inherent magnification factors, such as 2, 4 and 8, to maintain the manifold consistence. As a simple cascade mechanism, the inherent factors cannot ensure the optimal performance without the relationship between the manifold relevance and the down-sampling scale. In this paper, we explore the relevance with the groups of gradually down-sampled training sets and divide the scales into different classes for robust manifold consistence. And then, to map the optimal coefficients from LR spaces to HR target ones better, we introduce the weight-mapping neighbour embedding model. Qualitative and quantitative evaluations demonstrate that the enhancing of the manifold relevance can promote effective face hallucination. Based on the weight-mapping and scale clustering, our algorithm achieves better results compared with the state-of-the-art methods.