Search

A general deep learning (DL) mechanism for a multiple hidden layer feed-forward neural network contains two parts, i.e., 1) an unsupervised greedy layer-wise training and 2) a supervised fine-tuning which is usually an iterative process. Although this mechanism has been demonstrated in many fields to be able to significantly improve the generalization of neural network, there is no clear evidence to show which one of the two parts plays the essential role for the generalization improvement, resulting in an argument within the DL community. Focusing on this argument, this paper proposes a new DL approach to train multilayer feed-forward neural networks. This approach uses restricted Boltzmann machine (RBM) as the layer-wise training and uses the generalized inverse of a matrix as the supervised fine-tuning. Different from the general deep training mechanism like back-propagation (BP), the proposed approach does not need to iteratively tune the weights, and therefore, has many advantages such as quick training, better generalization, and high understandability, etc. Experimentally, the proposed approach demonstrates an excellent performance in comparison with BP-based DL and the traditional training method for multilayer random weight neural networks. To a great extent, this paper demonstrates that the supervised part plays a more important role than the unsupervised part in DL, which provides some new viewpoints for exploring the essence of DL. [ABSTRACT FROM AUTHOR]

This paper concerns the deployment problem of wireless sensor networks (WSNs) with mobile robotic sensor nodes for spatiotemporal monitoring. The proposed approach, deep reinforced learning tree (DRLT), utilizes deep reinforcement learning (DRL) to improve the efficiency of searching the most informative sampling locations. The parameterized sampling locations in an infinite horizon space are modeled according to their spatiotemporal correlations and subject to various constraints, including field estimation error and information gain. And the model-based information gain can be calculated efficiently over an infinite horizon. In this manner, the effectiveness of the sampling locations is learned through DRLT during the exploration by the robotic sensors. Then DRLT can instruct the robotic sensors to avoid unnecessary sampling locations in future iterations. Also, it is proved in this paper that the proposed algorithm is capable of searching for the near-optimal sampling locations effectively and guaranteeing a minimum field estimation error. Simulation based on national oceanic and atmospheric administration (NOAA) datasets is presented, which demonstrates the significant enhancements made by the proposed algorithm. Compared with the traditional approaches, such as the information theory-based greedy approach or random sampling, the proposed method shows a superior performance with regard to both estimation error and planning efficiency. [ABSTRACT FROM AUTHOR]

Stereo matching methods consist of matching cost computation and several post processing steps. Deep learning methods have greatly raised the accuracy of matching cost and achieved the lowest error rate on several public datasets. However, their generality capabilities are not the best due to potential overfitting, which is the common problem of supervised learning approaches. This paper proposes a convolutional neural network (CNN) based cost estimation method for computing the similarity of image patches. In consideration of accuracy and generalization capability, small size convolution kernels are chosen in the convolution layer and dropout in the decision layer is used for preventing overfitting. After obtaining stereo matching cost from the output of the CNN, several post-processing operations are adopted for disparity optimization, which includes semi-global matching in 1-D from different directions, a left-right consistency check, and the slanted plane smoothing method. The method is evaluated on KITTI 2012, KITTI 2015, and Middlebury stereo datasets and the experimental results on the KITTI benchmark demonstrate the competitive accuracy performance of the approach. Additionally, to test the generalization of the method, a series of extended crossover experiments are conducted in which the training samples and testing samples come from different datasets. The results indicate superior generalization capability of our method than other supervised learning methods. [ABSTRACT FROM AUTHOR]

Recommender systems (RSs) commonly adopt a user-item rating matrix to describe users’ preferences on items. With users and items exploding, such a matrix is usually high-dimensional and sparse (HiDS). Recently, the idea of deep learning has been applied to RSs. However, current deep-structured RSs suffer from high computational complexity. Enlightened by the idea of deep forest, this paper proposes a deep latent factor model (DLFM) for building a deep-structured RS on an HiDS matrix efficiently. Its main idea is to construct a deep-structured model by sequentially connecting multiple latent factor (LF) models instead of multilayered neural networks through a nonlinear activation function. Thus, the computational complexity grows linearly with its layer count, which is easy to resolve in practice. The experimental results on four HiDS matrices from industrial RSs demonstrate that when compared with state-of-the-art LF models and deep-structured RSs, DLFM can well balance the prediction accuracy and computational efficiency, which well fits the desire of industrial RSs for fast and right recommendations. [ABSTRACT FROM AUTHOR]

Convolutional neural networks (CNNs) have yielded state-of-the-art performance in image classification and other computer vision tasks. Their application in fire detection systems will substantially improve detection accuracy, which will eventually minimize fire disasters and reduce the ecological and social ramifications. However, the major concern with CNN-based fire detection systems is their implementation in real-world surveillance networks, due to their high memory and computational requirements for inference. In this paper, we propose an original, energy-friendly, and computationally efficient CNN architecture, inspired by the SqueezeNet architecture for fire detection, localization, and semantic understanding of the scene of the fire. It uses smaller convolutional kernels and contains no dense, fully connected layers, which helps keep the computational requirements to a minimum. Despite its low computational needs, the experimental results demonstrate that our proposed solution achieves accuracies that are comparable to other, more complex models, mainly due to its increased depth. Moreover, this paper shows how a tradeoff can be reached between fire detection accuracy and efficiency, by considering the specific characteristics of the problem of interest and the variety of fire data. [ABSTRACT FROM AUTHOR]

A system to perform video analytics is proposed using a dynamically tuned convolutional network. Videos are fetched from cloud storage, preprocessed, and a model for supporting classification is developed on these video streams using cloud-based infrastructure. A key focus in this paper is on tuning hyper-parameters associated with the deep learning algorithm used to construct the model. We further propose an automatic video object classification pipeline to validate the system. The mathematical model used to support hyper-parameter tuning improves performance of the proposed pipeline, and outcomes of various parameters on system’s performance is compared. Subsequently, the parameters that contribute toward the most optimal performance are selected for the video object classification pipeline. Our experiment-based validation reveals an accuracy and precision of 97% and 96%, respectively. The system proved to be scalable, robust, and customizable for a variety of different applications. [ABSTRACT FROM AUTHOR]

Fault diagnosis plays an important role in modern industry. With the development of smart manufacturing, the data-driven fault diagnosis becomes hot. However, traditional methods have two shortcomings: 1) their performances depend on the good design of handcrafted features of data, but it is difficult to predesign these features and 2) they work well under a general assumption: the training data and testing data should be drawn from the same distribution, but this assumption fails in many engineering applications. Since deep learning (DL) can extract the hierarchical representation features of raw data, and transfer learning provides a good way to perform a learning task on the different but related distribution datasets, deep transfer learning (DTL) has been developed for fault diagnosis. In this paper, a new DTL method is proposed. It uses a three-layer sparse auto-encoder to extract the features of raw data, and applies the maximum mean discrepancy term to minimizing the discrepancy penalty between the features from training data and testing data. The proposed DTL is tested on the famous motor bearing dataset from the Case Western Reserve University. The results show a good improvement, and DTL achieves higher prediction accuracies on most experiments than DL. The prediction accuracy of DTL, which is as high as 99.82%, is better than the results of other algorithms, including deep belief network, sparse filter, artificial neural network, support vector machine and some other traditional methods. What is more, two additional analytical experiments are conducted. The results show that a good unlabeled third dataset may be helpful to DTL, and a good linear relationship between the final prediction accuracies and their standard deviations have been observed. [ABSTRACT FROM AUTHOR]

In the area of community question answering (CQA), answer selection and answer ranking are two tasks which are applied to help users quickly access valuable answers. Existing solutions mainly exploit the syntactic or semantic correlation between a question and its related answers (Q&A), where the multifacet domain effects in CQA are still underexplored. In this paper, we propose a unified model, enhanced attentive recurrent neural network (EARNN), for both answer selection and answer ranking tasks by taking full advantages of both Q&A semantics and multifacet domain effects (i.e., topic effects and timeliness). Specifically, we develop a serialized long short-term memory to learn the unified representations of Q&A, where two attention mechanisms at either sentence level or word level are designed for capturing the deep effects of topics. Meanwhile, the emphasis of Q&A can be automatically distinguished. Furthermore, we design a time-sensitive ranking function to model the timeliness in CQA. To effectively train EARNN, a question-dependent pairwise learning strategy is also developed. Finally, we conduct extensive experiments on a real-world dataset from Quora. Experimental results validate the effectiveness and interpretability of our proposed EARNN model. [ABSTRACT FROM AUTHOR]

This paper presents a complete solution for the hardware design of a memristor-based long short-term memory (MLSTM) network. Throughout the design process, we fully consider the external and internal structures of the long short-term memory (LSTM), both of which are efficiently implemented by memristor crossbars. In the specific design of the internal structure, the parameter sharing mechanism is used between the LSTM cells to minimize the hardware design scale. In particular, we designed a circuit that requires only one memristor crossbar for each unit in the LSTM cell. The activation function, including sigmoid and tanh (hyperbolic tangent function), involved in each unit is approximated by a piecewise function, which is designed with the corresponding hardware. To verify the effectiveness of the system we designed, we test it on IMDB and SemEval datasets. Considering the huge impact of the dimensions of the input data on the scale of the hardware design, we use word2vector instead of one-hot encoding for the input data encoding. With the parameter sharing mechanism, the transformed vectors are input in different periods, so only 65 memristive crossbars are needed in the entire system to complete the sentiment analysis of the input text. The experimental results verify the effectiveness of our proposed MLSTM system. [ABSTRACT FROM AUTHOR]

Accurate knowledge of the vehicle states is the foundation of vehicle motion control. However, in real implementations, sensory signals are always corrupted by delays and noises. Network induced time-varying delays and measurement noises can be a hazard in the active safety of over-actuated electric vehicles (EVs). In this paper, a brain-inspired proprioceptive system based on state-of-the-art deep learning and data fusion technique is proposed to solve this problem in autonomous four-wheel actuated EVs. A deep recurrent neural network (RNN) is trained by the noisy and delayed measurement signals to make accurate predictions of the vehicle motion states. Then unscented Kalman predictor, which is the adaption of unscented Kalman filter in time-varying-delay situations, combines the predictions of the RNN and corrupted sensory signals to provide better perceptions of the locomotion. Simulations with a high-fidelity, CarSim, full-vehicle model are carried out to show the effectiveness of our RNN framework and the entire proprioceptive system. [ABSTRACT FROM AUTHOR]

Fuzzy modeling has many advantages over nonfuzzy methods, such as robustness with respect to uncertainties and less sensitivity to the varying dynamics of nonlinear systems. Data-driven fuzzy modeling needs to extract fuzzy rules from input and output data, and to train the fuzzy parameters of the fuzzy model. This paper takes advantages from deep learning, probability theory, fuzzy modeling, and extreme learning machines (ELMs). Restricted Boltzmann machine (RBM) and probability theory are used to overcome some common problems in data-driven modeling methods. The RBM is modified such that it can be trained with continuous values. A probability-based clustering method is proposed to partition the hidden features from the RBM. The obtained fuzzy rules have probability measurement. ELM and an optimization method are applied to train the fuzzy model. The proposed method is validated with two benchmark problems. [ABSTRACT FROM AUTHOR]

This paper proposes a novel multidomain features-based genetic algorithm (GA) optimized artificial immune system (AIS) framework for fault detection in real systems. Different from native real-valued negative selection algorithm (RNSA) that operates in original data space, this algorithm utilizes feature space transformation and diversity factor-based GA for optimized detector distribution in nonself feature space. The proposed framework comprises three stages namely; feature extraction, unsupervised feature selection, and GA optimized AIS. In the first stage, signal processing methods are applied to extract multidomain features (time-domain statistical, frequency domain statistical, and special features) of the system. In the second stage, two unsupervised methods namely, ${k}$ -NN clustering and pretraining using deep learning neural network are proposed for dominant fault-characterizing feature selection. Finally, in the third stage, the fault-characterizing feature vectors are used for system status categorization (i.e., normal, fault) using selected (fault-characterizing) features-based AIS method. The efficacy of the proposed framework is verified through experiments on motor bearing fault detection using vibration signal. The major accomplishment of the proposed combination of space transformation, feature selection and AIS (anomaly classification) techniques is the alleviation of computational burden on RNSA implementation. Moreover, GA optimized AIS fault diagnosis based on well-established features gives improved detection performance. [ABSTRACT FROM AUTHOR]

This paper presents a deep learning-based model reduction method for distributed parameter systems (DPSs). The proposed method includes three phases. In phase I, numerical or experimental data of the spatiotemporal distribution is reduced into low-dimensional representations using the deep auto-encoder (DAE). In phase II, the low-dimensional representations are used to establish the reduced-order model. In phase III, the reduced model is then used to reconstruct the high-dimensional DPS. Experimental studies are conducted to validate the proposed method. The proposed method is compared with the classical proper orthogonal decomposition method and demonstrates better modeling accuracy and efficiency in the experiments. [ABSTRACT FROM PUBLISHER]

Deep learning techniques have been widely used in autonomous driving community for the purpose of environment perception. Recently, it starts being adopted for learning end-to-end controllers for complex driving scenarios. However, the complexity and nonlinearity of the network architecture limits its interpretability to understand driving scenarios and judge the importance of certain visual regions in sensory scenes. In this paper, based on the convolutional neural network (CNN), we propose two complementary frameworks to automatically determine the most contributive regions of the input scenes, offering intuitive knowledge of how a trained end-to-end autonomous vehicle controller understands driving scenarios. In the first framework, a feature map-based method is proposed by leveraging current progress in CNN visualization, in which the deconvolution approach recovers the feature maps to extract features that contribute most to understand driving scenes. In the second framework, the importance level of regions is ranked using the error map between the labeled and predicted control inputs generated by occluding different parts of input scenes, thus providing a pixel-wise rank of importance. Test data sets with extracted contributive regions are input to the CNN controller. Then, different CNN controllers trained with the new data sets preprocessed using our proposed frameworks are verified via closed-loop tests. Results show that both the features identified from the first framework and the regions identified from the second framework are of crucial importance to scene understanding for the controller and can significantly affect the performance of CNN controllers. [ABSTRACT FROM AUTHOR]

Deep computation models achieve super performance for big data feature learning. However, training a deep computation model poses a significant challenge since a deep computation model typically involves a large number of parameters. Specially, it needs a high-performance computing server with a large-scale memory and a powerful computing unit to train a deep computation model, making it difficult to increase the size of a deep computation model further for big data feature learning on low-end devices such as conventional desktops and portable CPUs. In this paper, we propose an improved deep computation model based on the canonical polyadic decomposition scheme to compress the parameters and to improve the training efficiency. Furthermore, we devise a learning algorithm based on the back-propagation strategy to train the parameters of the proposed model. The learning algorithm can be directly performed on the compressed parameters to improve the training efficiency. Finally, we carry on the experiments on three representative datasets, i.e., CUAVE, SNAE2, and STL-10, to evaluate the performance of the proposed model by comparing with the conventional deep computation model and other two improved deep computation models based on the Tucker decomposition and the tensor-train network. Results demonstrate that the proposed model can compress parameters greatly and improve the training efficiency significantly with a low classification accuracy drop. [ABSTRACT FROM AUTHOR]

In the age of Internet of Things and Industrial 4.0, prognostic and health management (PHM) systems are used to collect massive real-time data from mechanical equipment. PHM big data has the characteristics of large-volume, diversity, and high-velocity. Effectively mining features from such data and accurately predicting the remaining useful life (RUL) of the rotating components with new advanced methods become issues in PHM. Traditional data driven prognostics is based on shallow learning architectures, requires establishing explicit model equations and much prior knowledge about signal processing techniques and prognostic expertise, and therefore is limited in the age of big data. This paper presents a deep learning-based approach for RUL prediction of rotating components with big data. The presented approach is tested and validated using data collected from a gear test rig and bearing run-to-failure tests and compared with existing PHM methods. The test results show the promising RUL prediction performance of the deep learning-based approach. [ABSTRACT FROM PUBLISHER]

X-ray screening systems have been used to safeguard environments in which access control is of paramount importance. Security checkpoints have been placed at the entrances to many public places to detect prohibited items, such as handguns and explosives. Generally, human operators are in charge of these tasks as automated recognition in baggage inspection is still far from perfect. Research and development on X-ray testing is, however, exploring new approaches based on computer vision that can be used to aid human operators. This paper attempts to make a contribution to the field of object recognition in X-ray testing by evaluating different computer vision strategies that have been proposed in the last years. We tested ten approaches. They are based on bag of words, sparse representations, deep learning, and classic pattern recognition schemes among others. For each method, we: 1) present a brief explanation; 2) show experimental results on the same database; and 3) provide concluding remarks discussing pros and cons of each method. In order to make fair comparisons, we define a common experimental protocol based on training, validation, and testing data (selected from the public \mathbb GDX ray database). The effectiveness of each method was tested in the recognition of three different threat objects: 1) handguns; 2) shuriken (ninja stars); and 3) razor blades. In our experiments, the highest recognition rate was achieved by methods based on visual vocabularies and deep features with more than 95% of accuracy. We strongly believe that it is possible to design an automated aid for the human inspection task using these computer vision algorithms. [ABSTRACT FROM AUTHOR]

The quadruped robot imitates the motions of four-legged animals with a superior flexibility and adaptability to complex terrains, compared with the wheeled and tracked robots. Its leader-following ability is unique to help a human to accomplish complex tasks in a more convenient way. However, long-term following is severely obstructed due to the high-frequency vibration of the quadruped robot and the unevenness of terrains. To solve this problem, a visual approach under a novel T-D-R framework is proposed. The proposed T-D-R framework is composed of a visual tracker based on correlation filter, a person detector with deep learning, and a person re-identification (re-ID) module. The result of the tracker is verified by the detector to improve tracking performance. Especially, the re-ID module is introduced to handle distractions and occlusion caused by other persons, where the convolutional correlation filter (CCF) is employed to discriminate the leader among multiple persons through recording the appearance information in the long run. By comparing the results of the tracker and the detector as well as their similarity scores with the leader identified by the re-ID module, a stable and real-time tracking of the leader can be guaranteed. Experiments reveal that our approach is effective in handling distractions, appearance changes, and illumination variations. A long-distance experiment on a quadruped robot indicates the validity of the proposed approach. [ABSTRACT FROM AUTHOR]

In lifelong machine learning (LML) systems, consecutive new tasks from changing circumstances are learned and added to the system. However, sufficiently labeled data are indispensable for extracting intertask relationships before transferring knowledge in classical supervised LML systems. Inadequate labels may deteriorate the performance due to the poor initial approximation. In order to extend the typical LML system, we propose a novel hierarchical lifelong learning algorithm (HLLA) consisting of two following layers: 1) the knowledge layer consisted of shared representations and integrated knowledge basis at the bottom and 2) parameterized hypothesis functions with features at the top. Unlabeled data is leveraged in HLLA for pretraining of the shared representations. We also have considered a selective inherited updating method to deal with intertask distribution shifting. Experiments show that our HLLA method outperforms many other recent LML algorithms, especially when dealing with higher dimensional, lower correlation, and fewer labeled data problems. [ABSTRACT FROM AUTHOR]

With the development of deep learning, semantic segmentation has received considerable attention within the robotics community. For semantic segmentation to be applied to mobile robots or autonomous vehicles, real-time processing is essential. In this article, a new real-time semantic segmentation network, called the adjacent feature propagation network (AFPNet), is proposed to achieve high performance and fast inference. AFPNet executes in real time on a commercial embedded GPU. The network includes two new modules. The local memory module (LMM) is the first; it improves the upsampling accuracy by propagating the high-level features to the adjacent grids. The cascaded pyramid pooling module (CPPM) is the second; it reduces computational time by changing the structure of the pyramid pooling module. Using these two modules, the proposed AFPNet achieved 76.4% mean intersection-over-union on the Cityscapes test dataset, outperforming other real-time semantic segmentation networks. Furthermore, AFPNet was successfully deployed on an embedded board Jetson AGX Xavier and applied to the real-world navigation of a mobile robot, proving that AFPNet can be effectively used in a variety of real-time applications. [ABSTRACT FROM AUTHOR]

Group recommendation has recently drawn a lot of attention to the recommender system community. Currently, several deep learning-based approaches are leveraged to learn preferences of groups for items and predict next items in which groups may be interested. Yet, their recommendation performance is still unsatisfactory due to sparse group–item interactions. To address this challenge, this study presents a novel model, called group recommendation model with two-stage deep learning (GRMTDL), which encompasses two sequential stages: 1) group representation learning (GRL) and 2) group preference learning (GPL). In GRL, we first construct an undirected tripartite graph over group–user–item interactions, and then employ it to accurately learn group semantic features through a spatial-based variational graph autoencoder network. While in GPL, we first introduce a dual PL-network that contains two structure-sharing subnetworks: 1) group PL-network employed for GPL and 2) user PL-network utilized for user preference learning. Then, we design a novel layered transfer learning (LTL) method to learn group preferences by alternately optimizing these two subnetworks. In particular, it can effectively absorb knowledge of user preferences into the process of GPL. Furthermore, extensive experiments on four real-world datasets demonstrate that the proposed GRMTDL model outperforms the state-of-the-art baselines for group recommendation. [ABSTRACT FROM AUTHOR]

Deep learning has been widely applied in many fields. Efficient optimization algorithms contribute a lot to the enhancement of deep learning. However, reverse studies on how deep learning can solve optimization problems, especially mathematical programming, are relatively scarce. In this work, we aim to initiate a discussion on using deep learning to solve parametric mathematical programming, which can be converted to find a mapping from parameter space to solution space. Given that deep convolutional neural networks (DCNNs) have been successfully applied in the field of computer vision, converting a mathematical programming problem into an image representation may provide the solution. This work takes the 0–1 knapsack problem (KP) as an example to build an original image representation method. After modifying the image details, the proposed method is extended to the general 0–1 linear programming (LP) problem, and a deep learning architecture is designed to solve the transferred computer vision problem. The efficiency of the proposed DCNNs method is validated on a large number of problems, and results show that this method can solve the 0–1 LP accurately and efficiently without iteration. The solution speed is 20 times faster than that of traditional optimization solvers. [ABSTRACT FROM AUTHOR]

Recent advances in deep learning have tremendously leveraged the performance of network representation learning (NRL). Multiple deep learning-based NRL models have been proposed recently to effectively handling primitive tasks of information network analysis and mining (INAM) domain, including link prediction (LP). LP is considered as an important one due to its multiple applications in many disciplines. In the recent few years, LP in dynamic networks has attracted a lot of attention from researchers to propose novel algorithms for better capturing both rich structural and evolutional information of complex information networks (INs). However, recent models are mainly concentrated on preserving the sequential representations of a given network over time. They have largely ignored other important structural features, such as: intracommunity which contributes to the creation of links between network nodes. In this article, we propose a novel community-driven dynamic NRL technique upon the RNN+GCN framework, called: ComGCN. Specifically, the ComGCN model is a combination of microscopic (node embedding-based) and mesoscopic (intracommunity-based) dynamic network embedding approach which enable effectively handling the LP problem in context of dynamism. Extensive experiments on real-world dynamic networks demonstrated the effectiveness of the proposed model compared with recent state-of-the-art baselines. [ABSTRACT FROM AUTHOR]

Predicting attention-modulated brain responses is a major area of investigation in brain–computer interface (BCI) research that aims to translate neural activities into useful control and communication commands. Such studies involve collecting electroencephalographic (EEG) data from subjects to train classifiers for decoding users’ mental states. However, various sources of inter or intrasubject variabilities in brain signals render training classifiers in BCI systems challenging. From a machine learning perspective, this model training generally follows a common methodology: 1) apply some type of feature extraction, which can be time-consuming and may require domain knowledge and 2) train a classifier using extracted features. The advent of deep learning technologies has offered unprecedented opportunities to not only construct remarkably accurate classifiers but also to integrate the feature extraction stage into the classifier construction. Although integrating feature extraction, which is generally domain-dependent, into the classifier construction is a considerable advantage of deep learning models, the process of architecture selection for BCIs generally depends on domain knowledge. In this study, we examine the feasibility of conducting a systematic model selection combined with mainstream deep learning architectures to construct accurate classifiers for decoding P300 event-related potentials. In particular, we present the results of 232 convolutional neural networks (CNNs) (4 datasets $\times58$ structures), 36 long short-term memory cells (LSTMs) (4 datasets $\times9$ structures), and 320 hybrid CNN-LSTM models (4 datasets $\times80$ structures) of varying complexity. Our empirical results show that in the classification of P300 waveforms, the constructed predictive models can outperform the current state-of-the-art deep learning architectures, which are partially or entirely inspired by domain knowledge. The source codes and constructed models are available at https://github.com/berdakh/P3Net. [ABSTRACT FROM AUTHOR]

Packet routing is one of the fundamental problems in computer networks in which a router determines the next-hop of each packet in the queue to get it as quickly as possible to its destination. Reinforcement learning (RL) has been introduced to design autonomous packet routing policies with local information of stochastic packet arrival and service. However, the curse of dimensionality of RL prohibits the more comprehensive representation of dynamic network states, thus limiting its potential benefit. In this article, we propose a novel packet routing framework based on multiagent deep RL (DRL) in which each router possess an independent long short term memory (LSTM) recurrent neural network (RNN) for training and decision making in a fully distributed environment. The LSTM RNN extracts routing features from rich information regarding backlogged packets and past actions, and effectively approximates the value function of Q-learning. We further allow each route to communicate periodically with direct neighbors so that a broader view of network state can be incorporated. The experimental results manifest that our multiagent DRL policy can strike the delicate balance between congestion-aware and shortest routes, and significantly reduce the packet delivery time in general network topologies compared with its counterparts. [ABSTRACT FROM AUTHOR]

Although great progress has been made in generic object detection by advanced deep learning techniques, detecting small objects from images is still a difficult and challenging problem in the field of computer vision due to the limited size, less appearance, and geometry cues, and the lack of large-scale datasets of small targets. Improving the performance of small object detection has a wider significance in many real-world applications, such as self-driving cars, unmanned aerial vehicles, and robotics. In this article, the first-ever survey of recent studies in deep learning-based small object detection is presented. Our review begins with a brief introduction of the four pillars for small object detection, including multiscale representation, contextual information, super-resolution, and region-proposal. Then, the collection of state-of-the-art datasets for small object detection is listed. The performance of different methods on these datasets is reported later. Moreover, the state-of-the-art small object detection networks are investigated along with a special focus on the differences and modifications to improve the detection performance comparing to generic object detection architectures. Finally, several promising directions and tasks for future work in small object detection are provided. Researchers can track up-to-date studies on this webpage available at: https://github.com/tjtum-chenlab/SmallObjectDetectionList. [ABSTRACT FROM AUTHOR]

Accurate prediction of solar energy is an important issue for photovoltaic power plants to enable early participation in energy auction industries and cost-effective resource planning. This article introduces a new deep learning-based multistep ahead approach to improve the forecasting performance of global horizontal irradiance (GHI). A deep convolutional long short-term memory is used to extract optimal features for accurate prediction of the GHI. The performance of such deep neural networks directly depends on their architectures. To deal with this problem, a swarm evolutionary optimization method, called the sine-cosine algorithm, is applied and advanced to automatically optimize the network architecture. A three-phase modification model is proposed to increase the diversity of population and avoid premature convergence in the optimization mechanism. The performance of the proposed method is investigated using three datasets collected from three solar stations in the east of the United States. The experimental results demonstrate the superiority of the proposed method in comparison to other forecasting models. [ABSTRACT FROM AUTHOR]

Deep reinforcement learning (RL) is achieving significant success in various applications like control, robotics, games, resource management, and scheduling. However, the important problem of emergency evacuation, which clearly could benefit from RL, has been largely unaddressed. Indeed, emergency evacuation is a complex task that is difficult to solve with RL. An emergency situation is highly dynamic, with a lot of changing variables and complex constraints that make it challenging to solve. Also, there is no standard benchmark environment available that can be used to train RL agents for evacuation. A realistic environment can be complex to design. In this article, we propose the first fire evacuation environment to train RL agents for evacuation planning. The environment is modeled as a graph capturing the building structure. It consists of realistic features like fire spread, uncertainty, and bottlenecks. The implementation of our environment is in the OpenAI gym format, to facilitate future research. We also propose a new RL approach that entails pretraining the network weights of a DQN-based agent [DQN/Double-DQN (DDQN)/Dueling-DQN] to incorporate information on the shortest path to the exit. We achieved this by using tabular $Q$ -learning to learn the shortest path on the building model’s graph. This information is transferred to the network by deliberately overfitting it on the $Q$ -matrix. Then, the pretrained DQN model is trained on the fire evacuation environment to generate the optimal evacuation path under time varying conditions due to fire spread, bottlenecks, and uncertainty. We perform comparisons of the proposed approach with state-of-the-art RL algorithms like DQN, DDQN, Dueling-DQN, PPO, VPG, state-action-reward-state-action (SARSA), actor–critic method, and ACKTR. The results show that our method is able to outperform state-of-the-art models by a huge margin including the original DQN-based models. Finally, our model is tested on a large and complex real building consisting of 91 rooms, with the possibility to move to any other room, hence giving 8281 actions. In order to reduce the action space, we propose a strategy that involves one step simulation. That is, an action importance vector is added to the final output of the pretrained DQN and acts like an attention mechanism. Using this strategy, the action space is reduced by 90.1%. In this manner, the model is able to deal with large action spaces. Hence, our model achieves near optimal performance on the real world emergency environment. [ABSTRACT FROM AUTHOR]

This article presents a simple and intuitive solution for multilabel image classification, which achieves the competitive performance on the popular COCO and PASCAL VOC benchmarks. The main idea is to capture how humans perform this task: we recognize both labels (i.e., objects and attributes) and the correlation of labels at the same time. Here, label recognition is performed by a standard ConvNet pipeline, whereas label correlation modeling is done by projecting both labels and image features extracted by the ConvNet to a common latent vector space. Specifically, we carefully design the loss function to ensure that: 1) labels and features that co-appear frequently are close to each other in the latent space and 2) conversely, labels/features that do not appear together are far apart. This information is then combined with the original ConvNet outputs to form the final prediction. The whole model is trained end-to-end, with no additional supervised information other than the image-level supervised information. Experiments show that the proposed method consistently outperforms previous approaches on COCO and PASCAL VOC in terms of mAP, macro/micro precision, recall, and $F$ -measure. Further, our model is highly efficient at test time, with only a small number of additional weights compared to the base model for direct label recognition. [ABSTRACT FROM AUTHOR]

Dynamic emotion is typically used to facilitate human–machine interactions. Conversational data from social media contain a considerable amount of useful information, and such data are the foundation for researching dynamic and artificial emotion. At present, most human–machine interaction systems focus on the complexity and accuracy of the dialog but neglect the emotional characteristics of the speaker. When generating a dialog considering the emotional personality of the interlocutor, controlling, and guiding the dialog to a specified direction are essential. This article presents a system for studying dynamic emotions in human-computer interaction from the perspective of emotional transfer and guidance. Based on the emotional state of the interlocutor and the distribution of emotional transfer, the process of emotional transfer is simulated and sampled, and the sequence of emotional guidance is generated. In this system, two algorithms are proposed. A generative Markov chain Monte Carlo (GEN-MCMC) algorithm is proposed to generate a variety of emotional transfer sequences that fit the talke’s personality dynamically based on the real-world dialog. Further, a guiding MCMC (GUI-MCMC) algorithm-based GEN-MCMC is proposed to generate the emotional guiding sequences. The generated emotional sequences by GEN-MCMC were evaluated in two aspects: 1) consistency and 2) diversity. The experimental results show that the GEN-MCMC algorithm performs better than the general sequence generation algorithm in terms of consistency and diversity in generating emotional states. The GUI-MCMC was able to generate a proper stimulus sequence when given the first and target emotions. An emotional stimulus sequence can simulate the emotional transfer of the interlocutor in the process of dialogue, and give the observer appropriate reference to guide and control the emotions of dialogue. The experimental results show that the proposed system can effectively model the dynamic emotion in emotional transfer and guidance, which can be further used to build chat robots, intelligent assistants, and human–machine interaction systems. The models can also be used for emotional induction and enhance the feel-good or feel-terrible factor in human–machine communication applications, such as medical treatment of mental diseases, interrogation, and psychological attack and defense. [ABSTRACT FROM AUTHOR]

The control of epidemics taking place in complex networks has been an increasingly active topic in public health management. In this article, we propose an efficient networked epidemic control system, where a modified susceptible-exposed-infected-vigilant (SEIV) model is first built to simulate epidemic spreading. Then, different from existing continuous resource models which abstractly map resources to parameters of epidemic models, a concrete resource description model is built to simulate real-world goods/services and their allocation. Based on the two models, a cost-constraint subset selection problem in epidemic control is identified. To solve the problem, a swarm-based stochastic optimization policy is proposed, where each particle in the swarm can determine its own solutions according to the guidance of its superior peers and historical searching experience of the whole swarm, without extra problem-relative information. Theoretical proof about system equilibrium is provided, which is consistent with experimental observations. The competitive performance of the proposed optimizer is validated by theoretical analysis and comparison experiments. Finally, an application case is provided to illustrate the practicability. [ABSTRACT FROM AUTHOR]

Fault diagnosis of rolling bearings is an essential process for improving the reliability and safety of the rotating machinery. It is always a major challenge to ensure fault diagnosis accuracy in particular under severe working conditions. In this article, a deep adversarial domain adaptation (DADA) model is proposed for rolling bearing fault diagnosis. This model constructs an adversarial adaptation network to solve the commonly encountered problem in numerous real applications: the source domain and the target domain are inconsistent in their distribution. First, a deep stack autoencoder (DSAE) is combined with representative feature learning for dimensionality reduction, and such a combination provides an unsupervised learning method to effectively acquire fault features. Meanwhile, domain adaptation and recognition classification are implemented using a Softmax classifier to augment classification accuracy. Second, the effects of the number of hidden layers in the stack autoencoder network, the number of neurons in each hidden layer, and the hyperparameters of the proposed fault diagnosis algorithm are analyzed. Third, comprehensive analysis is performed on real data to validate the performance of the proposed method; the experimental results demonstrate that the new method outperforms the existing machine learning and deep learning methods, in terms of classification accuracy and generalization ability. [ABSTRACT FROM AUTHOR]

The ever-increasing complexity of robot applications induces the need for methods to approach problems with no (viable) analytical solution. Deep learning (DL) provides a set of tools to address this kind of problems. This survey presents a categorization of the major challenges in robotics that leverage DL technologies and introduces representative examples of successful solutions for the described problems. We also consider the question when and whether to use modular, monolithic models or end-to-end DL, in order to provide a guideline for the selection of the correct model structure and training strategy. By doing so, the current role and adaptability of different techniques at different hierarchical levels of a robot-application can be highlighted, thus providing a well-structured basis to assist future approaches. [ABSTRACT FROM AUTHOR]

With the exponential growth of data and the high demand for the analysis of large datasets, the MapReduce framework has been widely utilized to process data in a timely, cost-effective manner. It is well-known that the performance of MapReduce is limited by its default configuration parameters, and there are a few research studies that have focused on finding the optimal configurations to improve the performance of the MapReduce framework. Recently, machine learning based approaches have been receiving more attention to be utilized to auto configure the MapReduce parameters to account for the dynamic nature of the applications. In this article, we propose and develop a reinforcement learning (RL)-based scheme, named RL-MRCONF, to automatically configure the MapReduce parameters. Specifically, we explore and experiment with two variations of RL-MRCONF; one variation is based on the traditional RL algorithm and the second is based on the deep RL algorithm. Results obtained from simulations show that the RL-MRCONF has the ability to successfully and effectively auto-configure the MapReduce parameters dynamically according to changes in job types and computing resources. Moreover, simulation results show our proposed RL-MRCONF scheme outperforms the traditional RL-based implementation. Using datasets provided by MR-Perf, simulation results show that our proposed scheme provides around 50% performance improvement in terms of execution time when compared with MapReduce using default settings. [ABSTRACT FROM AUTHOR]

In zero-shot learning, attributes play as a bridge from original images to class labels. Therefore, to achieve accurate zero-shot image classification, we mainly focus on improving attribute prediction accuracy by taking full advantage of prior information about attribute from two aspects. First, we present a new attribute classifier called deep attribute prediction (DeepAP) model by using supervised deep convolutional neural networks (DCNNs), where the attribute label information participates in the training of DCNNs. Unlike common DCNNs that are usually used to extract image features, the constructed DCNNs are used to directly predict attribute values from the original input images. Thus, the designed DeepAP model can serve as the mapping from low-level image features to high-level semantic attributes in the traditional direct attribute prediction (DAP) model. Second, another prior information about attribute, i.e., class-attribute matrix is used to mine the attribute-class correlation with sparse representation coefficients. Since the attribute-class correlation can reflect different contributions of attributes to classification, we use it to define attribute weights and incorporate the idea of weighted attributes into DeepAP to form the deep weighted attribute prediction (DWAP) model. Experiments on three real datasets show that DWAP outperforms the deep attribute network and DAP on attribute prediction and zero-shot image classification. [ABSTRACT FROM AUTHOR]

Model inversion attack (MIA) is a cyber threat with an increasing alert even for deep-learning-based recognition systems (DLRSs). By targeting a DLRS under a scenario of attacker access to the model structure and parameters, MIA generates a data clone for a certain targeted class label. To avoid the possible threats of such MIA-generated data clones, this research work proposes a social IoT approach to a collaborative cyber-defense among the online recognition systems (RSs) sharing the targeted class label. Since, the generation of an MIA-clone is by targeting an RS model and using its structure, parameters, and class labels output scores in an iterative optimization process, the generated clone is partially inherent to the targeted model. Thus, it is expected for an MIA-clone to show a different performance on a secondary RS wherein the same targeted class label is included. It is because, in the MIA generation of the clone, not only the targeted class label but also other class labels, and model parameters and structure affect the process, while the second model has just the targeted class label in common with the target model. Deploying the Social Internet of Recognition Systems (SIoRS), the proposed technique utilizes a collaborative recognition by SIoRC which plays the role of a complementary recognition besides the targeted RS. The recognition output by the targeted RS is further verified by the SIoRS complementary recognition result. To avoid the MIA-targeted data clones, the verification of recognition is by the log-likelihood ratio test between the targeted RS and the SIoRS complementary recognition confidence scores. The proposed technique is evaluated by statistical analysis on deep face RSs in 10000 Monte Carlo runs for each of the conventional, dc-generative adversarial network (GAN) and $\alpha $ -GAN integrated MIA techniques in targeting two different user identities. The $Z$ scores of the fitted normal distribution of the log-likelihood ratios indicate almost 100% detection rate of clones generated by conventional MIA and 95.23% and 86% of clones, respectively, generated by DC-GAN and $\alpha $ -GAN integrated deep MIA techniques. [ABSTRACT FROM AUTHOR]