Search

Heterogeneous information network (HIN) embedding aims at learning the low-dimensional representation of nodes while preserving structure and semantics in a HIN. Existing methods mainly focus on static networks, while a real HIN usually evolves over time with the addition (deletion) of multiple types of nodes and edges. Because even a tiny change can influence the whole structure and semantics, the conventional HIN embedding methods need to be retrained to get the updated embeddings, which is time-consuming and unrealistic. In this paper, we investigate the problem of dynamic HIN embedding and propose a novel Dynamic HIN Embedding model (DyHNE) with meta-path based proximity. Specifically, we introduce the meta-path based first- and second-order proximities to preserve structure and semantics in HINs. As the HIN evolves over time, we naturally capture changes with the perturbation of meta-path augmented adjacency matrices. Thereafter, we learn the node embeddings by solving generalized eigenvalue problem effectively and employ eigenvalue perturbation to derive the updated embeddings efficiently without retraining. Experiments show that DyHNE outperforms the state-of-the-arts in terms of effectiveness and efficiency. [ABSTRACT FROM AUTHOR]

Visual surveillance for autonomous power facility inspection is considered to bethe prominent field of study in the power industry. This research field completely focuses on either object detection or image fusion which lacks the overall consideration. By considering this, a single end-to-end object detection method by incorporating the image fusion named Fast Detection Fusion Network (FDFNet) is proposed in this paper to output qualitative fused images with detection results. The parameters in the FDFnet are greatly reduced by sharing the feature extraction network between image fusion and object detection tasks, due to which a huge reduction in computational complexity is achieved. On this basis, the object detection algorithm performance on various types of power facility images is compared and analyzed. For experimentation purposes, an IR (infrared) and VIS (visible) image acquisition system has also been designed. In addition, a dataset named CVPower with different sets of images for power facility fusion detection is constructed for this research field. Experimental results demonstrate that the proposed method can achieve the mAP of not less than 70%, process 2 frames per second, and produce high qualitative fused images. [ABSTRACT FROM AUTHOR]

Traditional algorithm lacks the ability to generalize across different application scenarios and how to set the stopping conditions of network iteration in Stochastic Configuration Network (SCN) is a problem. This paper proposes an event recognition system based on minimum redundancy, maximum relevance (mRMR) and a modified Stochastic Configuration Network (CWCs-SCN) for the fiber Bragg grating (FBG) sensor system. mRMR is used to analyze the importance of 14th time-frequency domain features in different application scenarios. After that, a modified SCN (CWCs-SCN) neural network is used to solve the stopping condition of the network iteration. The experiments demonstrated that the average recognition accuracy of four kind of human intrusion signal can reach to 96.7%, which is better than others traditional methods. In addition, a more compact neural network is obtained compared to SCN. To verify the generalization ability of the proposed model, it is applied to gear fault diagnosis with an accuracy of 96.3%. This imply that the proposed method in this paper has a wide application prospect. [ABSTRACT FROM AUTHOR]

This paper serves as a survey of recent advances in large margin training and its theoretical foundations, mostly for (nonlinear) deep neural networks (DNNs) that are probably the most prominent machine learning models for large-scale data in the community over the past decade. We generalize the formulation of classification margins from classical research to latest DNNs, summarize theoretical connections between the margin, network generalization, and robustness, and introduce recent efforts in enlarging the margins for DNNs comprehensively. Since the viewpoint of different methods is discrepant, we categorize them into groups for ease of comparison and discussion in the paper. Hopefully, our discussions and overview inspire new research work in the community that aim to improve the performance of DNNs, and we also point to directions where the large margin principle can be verified to provide theoretical evidence why certain regularizations for DNNs function well in practice. We managed to shorten the paper such that the crucial spirit of large margin learning and related methods are better emphasized. [ABSTRACT FROM AUTHOR]

In this paper, we present a comprehensive review of the imbalance problems in object detection. To analyze the problems in a systematic manner, we introduce a problem-based taxonomy. Following this taxonomy, we discuss each problem in depth and present a unifying yet critical perspective on the solutions in the literature. In addition, we identify major open issues regarding the existing imbalance problems as well as imbalance problems that have not been discussed before. Moreover, in order to keep our review up to date, we provide an accompanying webpage which catalogs papers addressing imbalance problems, according to our problem-based taxonomy. Researchers can track newer studies on this webpage available at: https://github.com/kemaloksuz/ObjectDetectionImbalance. [ABSTRACT FROM AUTHOR]

Graph-based clustering plays an important role in the clustering area. Recent studies about graph neural networks (GNN) have achieved impressive success on graph-type data. However, in general clustering tasks, the graph structure of data does not exist such that GNN can not be applied to clustering directly and the strategy to construct a graph is crucial for performance. Therefore, how to extend GNN into general clustering tasks is an attractive problem. In this paper, we propose a graph auto-encoder for general data clustering, AdaGAE, which constructs the graph adaptively according to the generative perspective of graphs. The adaptive process is designed to induce the model to exploit the high-level information behind data and utilize the non-euclidean structure sufficiently. Importantly, we find that the simple update of the graph will result in severe degeneration, which can be concluded as better reconstruction means worse update. We provide rigorous analysis theoretically and empirically. Then we further design a novel mechanism to avoid the collapse. Via extending the generative graph models to general type data, a graph auto-encoder with a novel decoder is devised and the weighted graphs can be also applied to GNN. AdaGAE performs well and stably in different scale and type datasets. Besides, it is insensitive to the initialization of parameters and requires no pretraining. [ABSTRACT FROM AUTHOR]

Digital watermarking technology and watermark attack methods are mutually reinforcing and complementary. Currently, traditional watermark attack methods are relatively mature, but these traditional attack methods will inevitably damage the visual quality of original images (OIs). Therefore, this paper proposes a covert attack method called residual dense based imperceptible watermark attack network (RD-IWAN). First, this paper designs a watermark attack residual dense network (WARDN) based on the residual dense network (RDN), which can effectively remove the watermark information in the middle and low frequency features of the watermarked image (WMI). Second, to improve the attack ability of the network, this paper innovatively proposes a progressive preprocessing method based on the information enhancement preprocessing method. Concurrently, to ensure the imperceptibility of this watermark attack method, a comprehensive loss function that combines the perceptual loss and mean square error loss (MSE) of OI and attacked watermarked image (AWMI) is designed in this study. Finally, attack experiments are designed and performed on watermarks with different embedding strengths and sizes. Experimental results show that, compared to traditional attack methods, the watermark attack method proposed in this paper exhibits stronger attack ability and higher imperceptibility. [ABSTRACT FROM AUTHOR]

With the emergence of new services, the complex optical network environment makes it more difficult to predict network anomalies. This paper proposes a multi-index anomaly prediction scheme with hybrid supervised/unsupervised deep learning for elastic optical networks. Aimed at complex optical network indicators, the scheme presents three phases to enhance the abnormal prediction. The scheme first selects the most influential indicators of anomaly label among the mass of network indicators by calculating the Spearman correlation coefficient. Then, considering the timeliness of network data, it predicts time series of different indicators to analyze future network conditions by using long short-term memory neural network. In order to improve the accuracy and efficiency of the anomaly detection model, the scheme further establishes a deep neural network for anomaly classification. We also discuss how to process data without anomaly labels. The feasibility of the proposed scheme is verified on a real network dataset. Experimental results show that the scheme can predict the occurrence of future network anomalies with high accuracy, protect network services from potential abnormalities, and enhance the stability and robustness of the network. [ABSTRACT FROM AUTHOR]

Visual classification algorithms based-on Deep Neural Networks (DNN) have been widely adopted in autonomous vehicle design. However, DNN suffers from adversarial attacks including pixel attacks and patch attacks, and its adoption may introduce new vulnerability into such security-critical scenarios. Existing defense techniques only focus on defending against one category, either pixel attacks or patch attacks, but does not translate to the other. Hence, the design of a practical comprehensive real-time defense algorithm for DNN-based classifiers presents a challenging task in this adversarial context. This paper attempts to address the abovementioned problem by combining Compressive Sensing with Generative neural networks (CSG) to construct an efficient defense framework, in conjunction with the proposal of a classifier-aware adversarial training way. Extensive experiments have been conducted using the LISA road sign dataset to evaluate the performance of CSG. The results show its superiority in comprehensively defending adversarial examples generated using attacks including CW-L2, FGSM and Sticker, compared with other state-of-the-art defense techniques. [ABSTRACT FROM AUTHOR]

This paper proposes to administer a multi-step artificial intelligence approach with an ensemble of adaptive neural networks (NNs) trained on 50000 samples to identify partial discharge (PD) diagnostic measurements for in-service medium voltage (MV) power cables. To evaluate the performance of the algorithm, a case study was performed on cables deliberately selected to contain both uncomplicated measurements and disruptive irregularities representative of conditions during field testing. Experimental test results prove that the proposed cable-specific adaptation improves PD identification accuracy, with further increment through the NN ensembles. The main contribution of the approach is in both the cable-specific adaption and the NN ensemble being applied to MV cable field measurements. [ABSTRACT FROM AUTHOR]

Over the last decade, deep neural networks (DNNs) are regarded as black-box methods, and their decisions are criticized for the lack of explainability. Existing attempts based on local explanations offer each input a visual saliency map, where the supporting features that contribute to the decision are emphasized with high relevance scores. In this paper, we improve the saliency map based on differentiated explanations, of which the saliency map not only distinguishes the supporting features from backgrounds but also shows the different degrees of importance of the various parts within the supporting features. To do this, we propose to learn a differentiated relevance estimator called DRE, where a carefully-designed distribution controller is introduced to guide the relevance scores towards right-skewed distributions. DRE can be directly optimized under pure classification losses, enabling higher faithfulness of explanations and avoiding non-trivial hyper-parameter tuning. The experimental results on three real-world datasets demonstrate that our differentiated explanations significantly improve the faithfulness with high explainability. Our code and trained models are available at https://github.com/fuweijie/DRE. [ABSTRACT FROM AUTHOR]

This paper proposes an innovative method to estimate the velocity of a moving body. This is achieved using solely raw data from a triad of low-cost inertial sensors, i.e. accelerometer and gyroscope, as well as a determined arrangement of magnetometer array. The proposed approach combines a magnetic field gradient-based Extended Kalman Filter (EKF), with a Bidirectional Long Short-Term Memory (BiLSTM) network. This is to better estimate the velocity, especially when the magnetic field disturbances are low, which causes other magnetic field-based methods to be inaccurate. The proposed method also makes it possible to well update the velocity regardless of sensor location, without any heavy computation or complex tuning, as the case for the Zero-Velocity Update Technique (ZUPT). The performance of the proposed approach is demonstrated through real experiments data using a Magneto-Inertial Tachymeter (MIT). The obtained results show the efficiency of the velocity estimation and possibly position, for different sensor placements and trajectory scenarios. [ABSTRACT FROM AUTHOR]

The performance of the recently proposed orientational beamforming (OBF) system for ultra-wideband (UWB) signals degrades seriously in multiuser situation. Although various techniques have been proposed in the literature to suppress multiuser interference in the time-hopping UWB system, there has been no research on interference suppression in the OBF UWB system. In this paper, we propose a nonlinear OBF system via a modified radial basis function (RBF) neural network, named the RBF-OBF system, to reject orientational UWB interferences caused by multiuser transmission in the OBF system. The RBF neural network is selected mainly because of its simple structure, convenient training process, and fast convergence speed. However, the conventional Euclidean-distance-based RBF neural network is not suitable for the training process used in this work, as the training samples are noiseless orientational steering vectors in frequency domain. Therefore, a modified RBF neural network is proposed in this paper, which evaluates the similarity between an input vector and the center vector of a hidden layer neuron by their signed complex correlation coefficient. Numerous simulations demonstrate that compared to the conventional OBF system, the proposed RBF-OBF system can significantly reduce the bit error rates (BERs) under different noise and interference situations. The BER reduction rate of the proposed RBF-OBF system is higher than 92% in additive white Gaussian noise channel and higher than 89% in line-of-sight multipath channel. [ABSTRACT FROM AUTHOR]

Academic research in the field of autonomous vehicles has reached high popularity in recent years related to several topics as sensor technologies, V2X communications, safety, security, decision making, control, and even legal and standardization rules. Besides classic control design approaches, Artificial Intelligence and Machine Learning methods are present in almost all of these fields. Another part of research focuses on different layers of Motion Planning, such as strategic decisions, trajectory planning, and control. A wide range of techniques in Machine Learning itself have been developed, and this article describes one of these fields, Deep Reinforcement Learning (DRL). The paper provides insight into the hierarchical motion planning problem and describes the basics of DRL. The main elements of designing such a system are the modeling of the environment, the modeling abstractions, the description of the state and the perception models, the appropriate rewarding, and the realization of the underlying neural network. The paper describes vehicle models, simulation possibilities and computational requirements. Strategic decisions on different layers and the observation models, e.g., continuous and discrete state representations, grid-based, and camera-based solutions are presented. The paper surveys the state-of-art solutions systematized by the different tasks and levels of autonomous driving, such as car-following, lane-keeping, trajectory following, merging, or driving in dense traffic. Finally, open questions and future challenges are discussed. [ABSTRACT FROM AUTHOR]

Local failure of rail track commonly grow from surface defects. Hence, timely detection of surface defects helps to identify potential hazards on the track and reduce the occurrence of railroad accidents. Since track surface defects are scattered and diverse, and different service life leads to different types of defects, it is crucial to detect surface defects in real time, with efficiency, reliability and robustness. To this end, a pixel-level defects segmentation method is proposed. In this paper, features are tessellated together at the Channel level to form denser features, allowing additional information on surface defects textures to be propagated among high-resolution layers. Dropout is performed on the weak correlations learned during the convolution, so that the convolution blocks can share a uniform weight matrix, reducing computation redundancy and model complexity. Firstly, the track datasets of different ages are categorized into four sets, then, the samples are normalized to grey scale by pre-processing, and fed into the proposed network for training. An evaluation of the proposed model on defective samples was performed to demonstrate the performance of the method, with an Accuracy of 97.47%, a Loss of 0.0061 and an Average Frame Rate of 0.033s. The results of different networks tested on the same dataset show that the proposed model exhibits strong stability, adaptability and robustness. In addition, the proposed model is assessed on two different datasets with distinct challenges, with Mean Intersection over Union yielding 2.13% and 3.77% boosts respectively. [ABSTRACT FROM AUTHOR]

Understanding human mobility patterns at the point-of-interest (POI) scale plays an important role in enhancing business intelligence in mobile environments. While large efforts have been made in this direction, most studies simply utilize POI check-ins to mine the concerned mobility patterns, the effectiveness of which is usually hindered due to data sparsity. To obtain better POI-based human mobility for mining, in this paper, we strive to directly annotate the POIs associated with raw user-generated mobility records. We propose a neural context fusion approach which integrates various context factors in people's POI-visiting behaviors. Our approach evaluates the preference and transition factors via representation learning. Notably, we incorporate an attention mechanism to deal with the randomized transitions in raw mobility. The domain knowledge factors, i.e., distance, time and popularity, remain effective and our approach further includes them from a data-driven perspective. Factors are automatically fused with a feed-forward neural network. Furthermore, we exploit a multi-head architecture to enhance the model expressiveness. Using two real-life data sets, we conduct our experimental study and find that our approach consistently outperforms the state-of-the-art baselines by at least 32 percent in accuracy. Besides, we demonstrate the utility of the obtained POI-based human mobility with a POI recommendation example. [ABSTRACT FROM AUTHOR]

This paper proposes a momentum algorithm based on Borges difference and an adaptive momentum (Adam) algorithm based on Borges difference to update parameters, which can adjust the momentum information more flexibly. The Borges difference is proposed from the definition of Borges derivative to be combined with the gradient algorithm in convolutional neural networks. The proposed momentum algorithm based on Borges difference and Adam algorithm based on Borges difference can be adjusted more flexibly in order to speed up the convergence. The parameter optimization algorithm with the Borges difference presents a better performance compared with the integer-order momentum algorithm and integer-order Adam algorithm, with the proposed nonlinear adjustment method for the parameter tuning of convolutional neural networks. By analyzing experimental results of Fashion-MNIST dataset and CIFAR-10 dataset, the optimization algorithms based on Borges difference proposed in this paper gain better effects on the optimization model compared with the corresponding ones based on the integer-order difference, and can speed up the convergence speed and recognition accuracy of the image recognition. [ABSTRACT FROM AUTHOR]

Bike-Sharing Systems (BSS), as a green and convenient transportation means, has attracted significant attention and developed rapidly around the world. However, rooted to the temporal-spatial variance of users’ demand, bike stations of BSS can easily run into the empty or full state, which undermines the performance of BSS and users’ experience. To solve this problem and rebalance the bikes efficiently, researchers have proposed lots of methods, especially from the operation research community. However, this problem is intrinsically an NP-hard problem, and most proposed methods cannot be applied to large-scale BSS. In this paper, inspired by recent advance in the AI area, we notice that for a specific bike-sharing system, similar rebalancing problem is solved every day. Thus, it is promising to learn useful knowledge from the past problem instances and use it in the future ones. In this paper, we adopt sequential to sequential learning technique for knowledge learning and use it for new problems. Evaluation on real-world dataset shows that our approach substantially outperforms existing rebalancing schemes. [ABSTRACT FROM AUTHOR]

Having stepped into the era of information explosion, storing, processing and analyzing the vast data sometimes are quite intractable problems. However, it is impossible for personal computer or devices to tackle with such heavy workloads. Then, companies that provides cloud computational services come into business. From the perspective of companies, the cost for providing fog computing services is much higher than the traditional computing services. Consequently, the price for real-time requests is more expensive than the reserved services. Aiming at minimizing the expenditures, the most important part is how many cloud services the customers should reserve in advance because different amounts they consume will yield different expenses and both of insufficient and excess consumption result in wastes. The emerging machine learning method provides a powerful tool to address such a prediction problem. In this paper, we propose a hierarchical Pythagorean fuzzy deep neural network (HPFDNN) to forecast the quantity of requisite cloud services. On account of obtaining the better interpretations of original data, beyond the employment of fuzzy logic, the neural representation is also utilized as a complementary method. The information or the knowledge acquired from fuzzy and neural perspectives are coalesced as the final transformed data to be put into the learning systems, so that the useful information concealed in the enormous contents can be effectively described. On the basis of the anticipation of the deep neural network, the consumers are able to decide the amount of cloud services to purchase. Numerical results based on the real data set from Carnegie Mellon University demonstrate that the proposed model yields the economical predictions and outperforms the prediction by the traditional deep neural network. [ABSTRACT FROM AUTHOR]

A novel method based on deep learning technique for the modal decomposition of the optical fields emerging from the few-mode fiber is demonstrated in this paper. By combining the advantages of principal component analysis (PCA) and Back-Propagation (BP) neural network, this scheme can reveal the exact superposition of eigenmodes. Firstly, PCA algorithm is applied to preprocess the target samples to reduce the computational complexity and extract the characteristics of the samples. Then, the mapping between the mode coefficients and the preprocessed near-field beam patterns is learned by using the BP neural network. The superiority of the proposed scheme is evaluated through simulation and experiment. The results show that the scheme can perform a complete modal decomposition within a few milliseconds, and it can still work well when the SNR level is as low as 20 dB. It is also worth noting that the method described in this work also has the advantages of short network training time, non-iterative, and low experimental equipment requirements. [ABSTRACT FROM AUTHOR]

Time series classification models have been garnering significant importance in the research community. However, not much research has been done on generating adversarial samples for these models. These adversarial samples can become a security concern. In this paper, we propose utilizing an adversarial transformation network (ATN) on a distilled model to attack various time series classification models. The proposed attack on the classification model utilizes a distilled model as a surrogate that mimics the behavior of the attacked classical time series classification models. Our proposed methodology is applied onto 1-nearest neighbor dynamic time warping (1-NN DTW) and a fully convolutional network (FCN), all of which are trained on 42 University of California Riverside (UCR) datasets. In this paper, we show both models were susceptible to attacks on all 42 datasets. When compared to Fast Gradient Sign Method, the proposed attack generates a larger faction of successful adversarial black-box attacks. A simple defense mechanism is successfully devised to reduce the fraction of successful adversarial samples. Finally, we recommend future researchers that develop time series classification models to incorporating adversarial data samples into their training data sets to improve resilience on adversarial samples. [ABSTRACT FROM AUTHOR]

It is widely acknowledged that the success of deep learning is built upon large-scale training data and tremendous computing power. However, the data and computing power are not always available for many real-world applications. In this paper, we address the machine learning problem where it lacks training data and limits computing power. Specifically, we investigate domain adaptation which is able to transfer knowledge from one labeled source domain to an unlabeled target domain, so that we do not need much training data from the target domain. At the same time, we consider the situation that the running environment is confined, e.g., in edge computing the end device has very limited running resources. Technically, we present the Faster Domain Adaptation (FDA) protocol and further report two paradigms of FDA: early stopping and amid skipping. The former accelerates domain adaptation by multiple early exit points. The latter speeds up the adaptation by wisely skip several amid neural network blocks. Extensive experiments on standard benchmarks verify that our method is able to achieve the comparable and even better accuracy but employ much less computing resources. To the best of our knowledge, there are very few works which investigated accelerating knowledge adaptation in the community. This work is expected to inspire the topic for more discussion. [ABSTRACT FROM AUTHOR]

Recent work has shown that convolutional networks can be substantially deeper, more accurate, and efficient to train if they contain shorter connections between layers close to the input and those close to the output. In this paper, we embrace this observation and introduce the Dense Convolutional Network (DenseNet), which connects each layer to every other layer in a feed-forward fashion. Whereas traditional convolutional networks with $L$ L layers have $L$ L connections—one between each layer and its subsequent layer—our network has $\frac{L(L+1)}{2}$ L (L + 1) 2 direct connections. For each layer, the feature-maps of all preceding layers are used as inputs, and its own feature-maps are used as inputs into all subsequent layers. DenseNets have several compelling advantages: they alleviate the vanishing-gradient problem, encourage feature reuse and substantially improve parameter efficiency. We evaluate our proposed architecture on four highly competitive object recognition benchmark tasks (CIFAR-10, CIFAR-100, SVHN, and ImageNet). DenseNets obtain significant improvements over the state-of-the-art on most of them, whilst requiring less parameters and computation to achieve high performance. [ABSTRACT FROM AUTHOR]

The growing number of Internet of Thing (IoT) and Ultra-Reliable Low Latency Communications (URLCC) use cases in next generation communication networks calls for the development of efficient Forward Error Correction (FEC) mechanisms. These use cases usually imply using short to mid-sized information blocks and requires low-complexity and/or fast decoding procedures. This paper investigates the joint learning of short to mid block-length coding schemes and associated Belief-Propagation (BP) like decoders using Machine Learning (ML) techniques. An interpretable auto-encoder (AE) architecture is proposed, ensuring scalability to block sizes currently challenging for ML-based linear block code design approaches. By optimizing a coding scheme w.r.t. the targeted decoder, the proposed system offers a good complexity/performance trade-off compared to various codes from literature with length up to 128 bits. [ABSTRACT FROM AUTHOR]

The conventional design of wireless communication systems typically relies on established mathematical models that capture the characteristics of different communication modules. Unfortunately, such design cannot be easily and directly applied to future wireless networks, which will be characterized by large-scale ultra-dense networks whose design complexity scales exponentially with the network size. Furthermore, such networks will vary dynamically in a significant way, which makes it intractable to develop comprehensive analytical models. Recently, deep learning-based approaches have emerged as potential alternatives for designing complex and dynamic wireless systems. However, existing learning-based methods have limited capabilities to scale with the problem size and to generalize with varying network settings. In this paper, we propose a scalable and generalizable neural calibration framework for future wireless system design, where a neural network is adopted to calibrate the input of conventional model-based algorithms. Specifically, the backbone of a traditional time-efficient algorithm is integrated with deep neural networks to achieve a high computational efficiency, while enjoying enhanced performance. The permutation equivariance property, carried out by the topological structure of wireless systems, is furthermore utilized to develop a generalizable neural network architecture. The proposed neural calibration framework is applied to solve challenging resource management problems in massive multiple-input multiple-output (MIMO) systems. Simulation results will show that the proposed neural calibration approach enjoys significantly improved scalability and generalization compared with the existing learning-based methods. [ABSTRACT FROM AUTHOR]

Image reconstruction for positron emission tomography (PET) is challenging because of the ill-conditioned tomographic problem and low counting statistics. Kernel methods address this challenge by using kernel representation to incorporate image prior information in the forward model of iterative PET image reconstruction. Existing kernel methods construct the kernels commonly using an empirical process, which may lead to unsatisfactory performance. In this paper, we describe the equivalence between the kernel representation and a trainable neural network model. A deep kernel method is then proposed by exploiting a deep neural network to enable automated learning of an improved kernel model and is directly applicable to single subjects in dynamic PET. The training process utilizes available image prior data to form a set of robust kernels in an optimized way rather than empirically. The results from computer simulations and a real patient dataset demonstrate that the proposed deep kernel method can outperform the existing kernel method and neural network method for dynamic PET image reconstruction. [ABSTRACT FROM AUTHOR]

Driven by the outstanding performance of neural networks in the structured euclidean domain, recent years have seen a surge of interest in developing neural networks for graphs and data supported on graphs. The graph is leveraged at each layer of the neural network as a parameterization to capture detail at the node level with a reduced number of parameters and computational complexity. Following this rationale, this paper puts forth a general framework that unifies state-of-the-art graph neural networks (GNNs) through the concept of EdgeNet. An EdgeNet is a GNN architecture that allows different nodes to use different parameters to weigh the information of different neighbors. By extrapolating this strategy to more iterations between neighboring nodes, the EdgeNet learns edge- and neighbor-dependent weights to capture local detail. This is a general linear and local operation that a node can perform and encompasses under one formulation all existing graph convolutional neural networks (GCNNs) as well as graph attention networks (GATs). In writing different GNN architectures with a common language, EdgeNets highlight specific architecture advantages and limitations, while providing guidelines to improve their capacity without compromising their local implementation. For instance, we show that GCNNs have a parameter sharing structure that induces permutation equivariance. This can be an advantage or a limitation, depending on the application. In cases where it is a limitation, we propose hybrid approaches and provide insights to develop several other solutions that promote parameter sharing without enforcing permutation equivariance. Another interesting conclusion is the unification of GCNNs and GATs —approaches that have been so far perceived as separate. In particular, we show that GATs are GCNNs on a graph that is learned from the features. This particularization opens the doors to develop alternative attention mechanisms for improving discriminatory power. [ABSTRACT FROM AUTHOR]

Several deep learning techniques have been used to detect pavement cracks for the partial replacement of inefficient traditional inspections. However, the extensively varying real-world situations limit the detection accuracy. While many existing studies utilized the attention modules in pavement crack detection to improve model performance, few studies considered the impact of “how” and “where” to add attention modules on model performance, namely the module optimization. Combined with the attention mechanism, a new pavement crack detection method was proposed based on the You Only Look Once 5th version (YOLOv5) in this paper. Considering two adding ways and three adding positions, the spatial and channel squeeze and excitation (SCSE) module and convolutional block attention module (CBAM) were used to build a total of 12 different attention models for the cracking detection. Each model was trained on 3248 images, and the weight with the best performance on the validation set was saved for testing. The test results show that the mAP@0.5:0.95 of the best attention model is improved by nearly 6.7% compared to the original model without the attention mechanism. In addition, it can process images at 13.15ms/pic while maintaining 94.4% precision, fully meeting the needs of real-time detection. Compared with the existing pavement crack detection methods, the advantages of the proposed method include a great detection speed, high accuracy, and good robustness. [ABSTRACT FROM AUTHOR]

With the boom of deep neural networks, blind image quality assessment (BIQA) has achieved great processes. However, the current BIQA metrics are limited when evaluating low-quality images as compared to medium-quality and high-quality images, which restricts their applications in real world problems. In this paper, we first identify that two challenges caused by distribution shift and long-tailed distribution lead to the compromised performance on low-quality images. Then, we propose an intermediary enhancement-based bilateral network with iterative training strategy for solving these two challenges. Drawing on the experience of transitive transfer learning, the proposed metric adaptively introduces enhanced intermediary images to transfer more information to low-quality images for mitigating the distribution shift. Our metric also adopts an iterative training strategy to deal with the long-tailed distribution. This strategy decouples feature extraction and score regression for better representation learning and regressor training. It not only transfers the knowledge learned from the earlier stage to the latter stage, but also makes the model pay more attention to long-tailed low-quality images. We conduct extensive experiments on five authentically distorted image quality datasets. The results show that our metric significantly improves the evaluating performance on low-quality images and delivers state-of-the-art intra-dataset results. During generalization tests, our metric also achieves the best cross-dataset performance. [ABSTRACT FROM AUTHOR]

mmWave communication requires accurate and continuous beam steering to overcome the severe propagation loss and user mobility. In this paper, we leverage a self-supervised deep learning approach to exploit sub-6 GHz channels and propose a novel method to predict beamforming vectors in the mmWave band for a single access point– user link. This complex channel-beam mapping is learned via data issued from the DeepMIMO dataset. We then compare our proposed method with existing supervised deep learning and classic reinforcement learning methods. Our simulations show that choosing an appropriate beam steering method depends on the target application and is a tradeoff between data rate and computational complexity. We also investigate tuning the size of our neural network depending on the number of transmit and receive antennas at the access point. Finally, we extend our method to the case of multiple links and introduce a federated learning (FL) approach to efficiently predict their mmWave beams by sharing only the weights of the locally trained neural networks (and not the local data). We investigate both synchronous and asynchronous FL methods. Our numerical simulations show the high potential of our approach, especially when the local available data is scarce or imperfect. [ABSTRACT FROM AUTHOR]

With the prevalence of embedded GPUs on mobile devices, power-efficient rendering has become a widespread concern for graphics applications. Reducing the power consumption of rendering applications is critical for extending battery life. In this paper, we present a new real-time power-budget rendering system to meet this need by selecting the optimal rendering settings that maximize visual quality for each frame under a given power budget. Our method utilizes two independent neural networks trained entirely by synthesized datasets to predict power consumption and image quality under various workloads. This approach spares time-consuming precomputation or runtime periodic refitting and additional error computation. We evaluate the performance of the proposed framework on different platforms, two desktop PCs and two smartphones. Results show that compared to the previous state of the art, our system has less overhead and better flexibility. Existing rendering engines can integrate our system with negligible costs. [ABSTRACT FROM AUTHOR]

Reinforcement learning (RL) has demonstrated its ability to solve high dimensional tasks by leveraging non-linear function approximators. However, these successes are mostly achieved by ’black-box’ policies in simulated domains. When deploying RL to the real world, several concerns regarding the use of a ’black-box’ policy might be raised. In order to make the learned policies more transparent, we propose in this paper a policy iteration scheme that retains a complex function approximator for its internal value predictions but constrains the policy to have a concise, hierarchical, and human-readable structure, based on a mixture of interpretable experts. Each expert selects a primitive action according to a distance to a prototypical state. A key design decision to keep such experts interpretable is to select the prototypical states from trajectory data. The main technical contribution of the paper is to address the challenges introduced by this non-differentiable prototypical state selection procedure. Experimentally, we show that our proposed algorithm can learn compelling policies on continuous action deep RL benchmarks, matching the performance of neural network based policies, but returning policies that are more amenable to human inspection than neural network or linear-in-feature policies. [ABSTRACT FROM AUTHOR]

This paper tackles the problem of training a deep convolutional neural network of both low-bitwidth weights and activations. Optimizing a low-precision network is very challenging due to the non-differentiability of the quantizer, which may result in substantial accuracy loss. To address this, we propose three practical approaches, including (i) progressive quantization; (ii) stochastic precision; and (iii) joint knowledge distillation to improve the network training. First, for progressive quantization, we propose two schemes to progressively find good local minima. Specifically, we propose to first optimize a network with quantized weights and subsequently quantize activations. This is in contrast to the traditional methods which optimize them simultaneously. Furthermore, we propose a second progressive quantization scheme which gradually decreases the bitwidth from high-precision to low-precision during training. Second, to alleviate the excessive training burden due to the multi-round training stages, we further propose a one-stage stochastic precision strategy to randomly sample and quantize sub-networks while keeping other parts in full-precision. Finally, we adopt a novel learning scheme to jointly train a full-precision model alongside the low-precision one. By doing so, the full-precision model provides hints to guide the low-precision model training and significantly improves the performance of the low-precision network. Extensive experiments on various datasets (e.g., CIFAR-100, ImageNet) show the effectiveness of the proposed methods. [ABSTRACT FROM AUTHOR]

In an aerial hybrid massive multiple-input multiple-output (MIMO) and orthogonal frequency division multiplexing (OFDM) system, how to design a spectral-efficient broadband multi-user hybrid beamforming with a limited pilot and feedback overhead is challenging. To this end, by modeling the key transmission modules as an end-to-end (E2E) neural network, this paper proposes a data-driven deep learning (DL)-based unified hybrid beamforming framework for both the time division duplex (TDD) and frequency division duplex (FDD) systems with implicit channel state information (CSI). For TDD systems, the proposed DL-based approach jointly models the uplink pilot combining and downlink hybrid beamforming modules as an E2E neural network. While for FDD systems, we jointly model the downlink pilot transmission, uplink CSI feedback, and downlink hybrid beamforming modules as an E2E neural network. Different from conventional approaches separately processing different modules, the proposed solution simultaneously optimizes all modules with the sum rate as the optimization object. Therefore, by perceiving the inherent property of air-to-ground massive MIMO-OFDM channel samples, the DL-based E2E neural network can establish the mapping function from the channel to the beamformer, so that the explicit channel reconstruction can be avoided with reduced pilot and feedback overhead. Besides, practical low-resolution phase shifters (PSs) introduce the quantization constraint, leading to the intractable gradient backpropagation when training the neural network. To mitigate the performance loss caused by the phase quantization error, we adopt the transfer learning strategy to further fine-tune the E2E neural network based on a pre-trained network that assumes the ideal infinite-resolution PSs. Numerical results show that our DL-based schemes have considerable advantages over state-of-the-art schemes. [ABSTRACT FROM AUTHOR]

Trajectory planning is essential for self-driving vehicles and has stringent requirements for accuracy and efficiency. The existing trajectory planning methods have limitations in the feasibility of planned trajectories and computational efficiency. This paper proposes a life-long learning framework to achieve effective and high-accuracy direct trajectory planning (DTP) tasks. Based on generative adversarial networks (GANs), this study develops a lightweight GDTP model to map the initial/final states and the control action sequence. Additionally, by embedding the GDTP into the rapidly-exploring random tree (RRT), a GDTP-RRT algorithm is further designed for long-distance and multi-stage planning tasks. Taking the tractor-trailer as an application case, we test the proposed method in multiple scenarios with varying characteristics. The experimental results show that the method can plan highly feasible trajectories in a short time, compared with the most applied algorithm – the cubic curve RRT* (CCRRT*). It is found that the tracking errors of our method are 29.1% and 44.1% lower than the CCRRT* in terms of position and heading angle. This paper provides an effective and stable vehicle trajectory planning method for complex self-driving tasks. [ABSTRACT FROM AUTHOR]

This paper presents an event-triggered model predictive control (MPC) strategy with learning terminal cost for robotic manipulators containing model uncertainty and input constraints. In the proposed MPC structure, an adaptive predictive model for the robotic system is established by radial basis function neural networks (RBFNNs) firstly. Then, a terminal cost adjusted by the global learning mechanism is constructed. Both global steady-state optimization and transient fast convergence are achieved by adding the learning terminal cost to the MPC scheme. After that, a triggering condition of the MPC solving is developed based on the predictive model’s weights and the predictive tracking error. Besides, the condition to avoid Zeno behavior is obtained. The recursive feasibility of the proposed MPC strategy is verified, and the ultimately uniformly boundedness (UUB) of all variables is proved according to the Lyapunov theorem. Finally, experiments based on an xMate7 Pro robot are conducted to demonstrate the effectiveness of the presented method. Note to Practitioners—The tracking control of robotic manipulators is a common and important problem in industrial applications, such as grasping, loading, unloading, et al. There exist some limitations in existing control methods. For example, general control strategies such as adaptive control, sliding mode control neglect the balance between control costs and expected performance; existing optimal control approaches rarely consider global steady-state optimization and transient fast convergence under the influence of model uncertainty. This paper is motivated by these limitations of robotic control design and inspired by model predictive control and optimal control theory. It develops a novel strategy for tracking control of robotic manipulators, which includes three following items: (1) adopt an approximation model constructed by neural networks as the predictive model for estimating the robotic system; (2) introduce a critic network into the terminal cost of the objective function of MPC for learning global optimal solution; (3) establish an event-triggered mechanism for solving of the optimization problem. The proposed method is verified by experimental results. In our future work, the proposed control strategy will be applied to more industrial processes. [ABSTRACT FROM AUTHOR]

In this paper, we introduce a structure-based neural network architecture, namely RC-Struct, for MIMO-OFDM symbol detection. The RC-Struct exploits the temporal structure of the MIMO-OFDM signals through reservoir computing (RC). A binary classifier leverages the repetitive constellation structure in the system to perform multi-class detection. The incorporation of RC allows the RC-Struct to be learned in a purely online fashion with extremely limited pilot symbols in each OFDM subframe. The binary classifier enables the efficient utilization of the precious online training symbols and allows an easy extension to high-order modulations without a substantial increase in complexity. Experiments show that the introduced RC-Struct outperforms both the conventional model-based symbol detection approaches and the state-of-the-art learning-based strategies in terms of bit error rate (BER). The advantages of RC-Struct over existing methods become more significant when rank and link adaptation are adopted. The introduced RC-Struct sheds light on combining communication domain knowledge and learning-based receive processing for 5G/5G-Advanced and Beyond. [ABSTRACT FROM AUTHOR]

Accurate prediction of the remaining useful life (RUL) of mechanical equipment has brought benefits of the company’s reasonable maintenance. However, in real industrial applications, owing to the change of working conditions and the interference of environment noise, it is of great difficulty to extract useful features from the collected signals, making it quite challenging to achieve high-precision RUL prediction of mechanical equipment. In order to overcome these issues, a new high-precision prediction method based on improved convolutional neural network (ICNN), residual attention mechanism with soft thresholding and gated recurrent unit (GRU) is proposed in this paper. To begin with, this method solves the limitations of manual feature extraction and makes the extracted feature representation more obvious by integrating one-dimensional depth separable convolution neural network and two-dimensional transpose convolution neural network. Then, soft thresholding and residual connection is inserted into the attention mechanism to help improve the prediction performance of RUL in noisy environment, making the prediction model adaptively set different thresholds for each sample according to its conditions and characteristics. Finally, the effectiveness of the proposed approach is verified by simulating turbofan engine and IEEE phm2010 data set, indicating that our proposed method has better performance than other approaches of literatures in prediction accuracy and time cost. [ABSTRACT FROM AUTHOR]

Point cloud completion aims at predicting a complete 3D shape from an incomplete input. It has important applications in the fields of intelligent manufacturing, augmented reality, virtual reality, self-driving cars, and intelligent robotics. Although deep learning-based point cloud completion technology has developed rapidly in recent years, there are still unsolved problems. Previous approaches predict each point independently and ignore contextual information. And, they usually predict a complete 3D shape based on a global feature vector extracted from an incomplete input, which leads to missing of some fine-grained details. In this paper, motivated by the transposed convolution and the “UNet” structure in neural networks for image processing, we propose a context-aware deep network termed as PCUNet for coarse-to-fine point cloud completion. It adopts an encoder-decoder structure, in which the encoder follows the design of the relation-shape convolutional neural network (RS-CNN), and the decoder consists of fully-connected layers and two stacked decoder modules for predicting complete point clouds. The contributions are twofold. First, we design the decoder module as a coordinate-guided context-aware upsampling module, in which contextual information can be taken into full account by neighbor aggregation. Second, to preserve fine-grained details in the input, we propose attention-enhanced skip connections for effective information propagation from the encoder to the decoder. Experiments are conducted on the widely used PCN and KITTI datasets. The results show that our proposed approach achieves competitive performance compared to the existing state-of-the-art approaches in terms of the Chamfer distance and the computational complexity metrics. [ABSTRACT FROM AUTHOR]

Factorization Machines (FMs) are effective in incorporating side information to overcome the cold-start and data sparsity problems in recommender systems. Traditional FMs adopt the inner product to model the second-order interactions between different attributes, which are represented via feature vectors. The problem is that the inner product violates the triangle inequality property of feature vectors. As a result, it cannot well capture fine-grained attribute interactions, resulting in sub-optimal performance. Recently, the euclidean distance is exploited in FMs to replace the inner product and has delivered better performance. However, previous FM methods including the ones equipped with the euclidean distance all focus on the attribute-level interaction modeling, ignoring the critical intrinsic feature correlations inside attributes. Thereby, they fail to model the complex and rich interactions exhibited in the real-world data. To tackle this problem, in this paper, we propose a FM framework equipped with generalized metric learning techniques to better capture these feature correlations. In particular, based on this framework, we present a Mahalanobis distance and a deep neural network (DNN) methods, which can effectively model the linear and non-linear correlations between features, respectively. Besides, we design an efficient approach for simplifying the model functions. Experiments on several benchmark datasets demonstrate that our proposed framework outperforms several state-of-the-art baselines by a large margin. Moreover, we collect a new large-scale dataset on second-hand trading to justify the effectiveness of our method over cold-start and data sparsity problems in recommender systems. [ABSTRACT FROM AUTHOR]

Deep convolutional neural networks (DCNNs) are currently the method of choice both for generative, as well as for discriminative learning in computer vision and machine learning. The success of DCNNs can be attributed to the careful selection of their building blocks (e.g., residual blocks, rectifiers, sophisticated normalization schemes, to mention but a few). In this paper, we propose $\Pi$ Π -Nets, a new class of function approximators based on polynomial expansions. $\Pi$ Π -Nets are polynomial neural networks, i.e., the output is a high-order polynomial of the input. The unknown parameters, which are naturally represented by high-order tensors, are estimated through a collective tensor factorization with factors sharing. We introduce three tensor decompositions that significantly reduce the number of parameters and show how they can be efficiently implemented by hierarchical neural networks. We empirically demonstrate that $\Pi$ Π -Nets are very expressive and they even produce good results without the use of non-linear activation functions in a large battery of tasks and signals, i.e., images, graphs, and audio. When used in conjunction with activation functions, $\Pi$ Π -Nets produce state-of-the-art results in three challenging tasks, i.e., image generation, face verification and 3D mesh representation learning. The source code is available at https://github.com/grigorisg9gr/polynomial_nets. [ABSTRACT FROM AUTHOR]

Point cloud registration is widely used in the research of intelligent vehicles, typical problems include map matching, visual odometer, pose estimation, etc. This paper proposes a deep learning-based registration method that can input point clouds directly, thereby preventing information loss of preprocessing needed by alternative deep-learning approaches. Our network, named DPFNet (Direct Point Feature Net), gradually downsamples the point cloud and aggregates points around determined reference points to formulate local features automatically. This is facilitated by a novel convolution-like operator and a novel loss function. The points in the point cloud are mapped to a high dimensional embedding through the designed deep neural network, where every embedding reflects the local feature of a specific spatial area. Based on the embedding features, correspondences between points can be estimated robustly and the registration between the point clouds can be obtained using an external geometric optimization algorithm. Experimental results on open benchmarks validate the proposed method and show that its performance is favourable over several baseline methods. Specifically, we test the proposed algorithm on KITTI benchmark, which shows its potential in tasks of intelligent vehicles, e.g., map matching, visual or LiDAR odometer. [ABSTRACT FROM AUTHOR]

As a challenging attempt to replace a traditional data structure with a learned model, this paper proposes a learned functional Bloom filter (L-FBF) for a key–value storage. The learned model in the proposed L-FBF learns the characteristics and the distribution of given data and classifies each input. It is shown through theoretical analysis that the L-FBF provides a lower search failure rate than a single FBF in the same memory size, while providing the same semantic guarantees. For model training, character-level neural networks are used with pretrained embeddings. In experiments, four types of different character-level neural networks are trained: a single gated recurrent unit (GRU), two GRUs, a single long short-term memory (LSTM), and a single one-dimensional convolutional neural network (1D-CNN). Experimental results prove the validity of theoretical results, and show that the L-FBF reduces the search failures by 82.8% to 83.9% when compared with a single FBF under the same amount of memory used. [ABSTRACT FROM AUTHOR]

This paper explores criticality-based real-time scheduling of neural-network-based machine inference pipelines in cyber-physical systems (CPS) to mitigate the effect of algorithmic priority inversion. We specifically focus on the perception subsystem, an important subsystem feeding other components (e.g., planning and control). In general, priority inversion occurs in real-time systems when computations that are of lower priority are performed together with or ahead of those that are of higher priority. In current machine perception software, significant priority inversion occurs because resource allocation to the underlying neural network models does not differentiate between critical and less critical data within a scene. To remedy this problem, in recent work, we proposed an architecture to partition the input data into regions of different criticality, then formulated a utility-based optimization problem to batch and schedule their processing in a manner that maximizes confidence in perception results, subject to criticality-based time constraints. This journal extension matures the work in several directions: (i) We extend confidence maximization to a generalized utility optimization formulation that accounts for criticality in the utility function itself, offering finer-grained control over resource allocation within the perception pipeline; (ii) we further instantiate and compare two different criticality metrics (distance-based and relative velocity-based) to understand their relative advantages; and (iii) we explore the limitations of the approach, specifically how inaccuracies in criticality-based attention cueing affect performance. All experiments are conducted on the NVIDIA Jetson AGX Xavier platform with a real-world driving dataset. [ABSTRACT FROM AUTHOR]

For deep learning, size is power. Massive neural nets trained on broad data for a spectrum of tasks are at the forefront of artificial intelligence. These large pre-trained models or "Jacks of All Trades" (JATs), when fine-tuned for downstream tasks, are gaining importance in driving deep learning advancements. However, environments with tight resource constraints, changing objectives and intentions, or varied task requirements, could limit the real-world utility of a singular JAT. Hence, in tandem with current trends towards building increasingly large JATs, this paper conducts an initial exploration into concepts underlying the creation of a diverse set of compact machine learning model sets. Composed of many smaller and specialized models, the Set of Sets is formulated to simultaneously fulfil many task settings and environmental conditions. A means to arrive at such a set tractably in one pass of a neuroevolutionary multitasking algorithm is presented for the first time, bringing us closer to models that are collectively "Masters of All Trades". [ABSTRACT FROM AUTHOR]