Search

Long short-term memory (LSTM) is a type of powerful deep neural network that has been widely used in many sequence analysis and modeling applications. However, the large model size problem of LSTM networks make their practical deployment still very challenging, especially for the video recognition tasks that require high-dimensional input data. Aiming to overcome this limitation and fully unlock the potentials of LSTM models, in this paper we propose to perform algorithm and hardware co-design towards high-performance energy-efficient LSTM networks. At algorithm level, we propose to develop fully decomposed hierarchical Tucker (FDHT) structure-based LSTM, namely FDHT-LSTM, which enjoys ultra-low model complexity while still achieving high accuracy. In order to fully reap such attractive algorithmic benefit, we further develop the corresponding customized hardware architecture to support the efficient execution of the proposed FDHT-LSTM model. With the delicate design of memory access scheme, the complicated matrix transformation can be efficiently supported by the underlying hardware without any access conflict in an on-the-fly way. Our evaluation results show that both the proposed ultra-compact FDHT-LSTM models and the corresponding hardware accelerator achieve very high performance. Compared with the state-of-the-art compressed LSTM models, FDHT-LSTM enjoys both order-of-magnitude reduction (more than $1000 \times$ 1000 × ) in model size and significant accuracy improvement (0.6% to 12.7%) across different video recognition datasets. Meanwhile, compared with the state-of-the-art tensor decomposed model-oriented hardware TIE, our proposed FDHT-LSTM architecture achieve $2.5\times$ 2. 5 × , $1.46\times$ 1. 46 × and $2.41\times$ 2. 41 × increase in throughput, area efficiency and energy efficiency, respectively on LSTM-Youtube workload. For LSTM-UCF workload, our proposed design also outperforms TIE with $1.9\times$ 1. 9 × higher throughput, $1.83\times$ 1. 83 × higher energy efficiency and comparable area efficiency. [ABSTRACT FROM AUTHOR]

This paper investigates the stochastic optimization problem focusing on developing scalable parallel algorithms for deep learning tasks. Our solution involves a reformation of the objective function for stochastic optimization in neural network models, along with a novel parallel computing strategy, coined the weighted aggregating stochastic gradient descent (WASGD). Following a theoretical analysis on the characteristics of the new objective function, WASGD introduces a decentralized weighted aggregating scheme based on the performance of local workers. Without any center variable, the new method automatically gauges the importance of local workers and accepts them by their contributions. Furthermore, we have developed an enhanced version of the method, WASGD+, by (1) implementing a designed sample order and (2) upgrading the weight evaluation function. To validate the new method, we benchmark our pipeline against several popular algorithms including the state-of-the-art deep neural network classifier training techniques (e.g., elastic averaging SGD). Comprehensive validation studies have been conducted on four classic datasets: CIFAR-100, CIFAR-10, Fashion-MNIST, and MNIST. Subsequent results have firmly validated the superiority of the WASGD scheme in accelerating the training of deep architecture. Better still, the enhanced version, WASGD+, is shown to be a significant improvement over its prototype. [ABSTRACT FROM AUTHOR]

Triangles are the basic substructure of networks and triangle counting (TC) has been a fundamental graph computing problem in numerous fields such as social network analysis. Nevertheless, like other graph computing problems, due to the high memory-computation ratio and random memory access pattern, TC involves a large amount of data transfers thus suffers from the bandwidth bottleneck in the traditional Von-Neumann architecture. To overcome this challenge, in this paper, we propose to accelerate TC with the emerging processing-in-memory (PIM) architecture through an algorithm-architecture co-optimization manner. To enable the efficient in-memory implementations, we come up to reformulate TC with bitwise logic operations (such as AND), and develop customized graph compression and mapping techniques for efficient data flow management. With the emerging computational Spin-Transfer Torque Magnetic RAM (STT-MRAM) array, which is one of the most promising PIM enabling techniques, the device-to-architecture co-simulation results demonstrate that the proposed TC in-memory accelerator outperforms the state-of-the-art GPU and FPGA accelerations by $12.2\times$ 12. 2 × and $31.8\times$ 31. 8 × , respectively, and achieves a $34\times$ 34 × energy efficiency improvement over the FPGA accelerator. [ABSTRACT FROM AUTHOR]

Finding top- $k$ k frequent items has been a hot issue in databases. Finding top- $k$ k persistent items is a new issue, and has attracted increasing attention in recent years. In practice, users often want to know which items are significant, i.e., not only frequent but also persistent. No prior art can address both of the above two issues at the same time. Also, for high-speed data streams, prior art cannot achieve high accuracy when the memory is tight. In this paper, we define a new issue, named finding significant items, and propose a novel algorithm namely LTC to address this issue. It includes two key techniques, Long-tail Restoring and CLOCK, as well as three optimizations. In addition, LTC is extended to support finding significant items with thresholds. We theoretically derive the correct rate and error bound, and conduct extensive experiments on three real datasets to test the performance of LTC. Our experimental results show that LTC can achieve $10^5$ 10 5 times higher accuracy in terms of average relative error than other related algorithms. Lastly, LTC is applied to a DDoS detection task and it shows that finding significant items is more powerful than finding frequent items. [ABSTRACT FROM AUTHOR]

In this paper, we study the formalism of unsupervised multi-class domain adaptation (multi-class UDA), which underlies a few recent algorithms whose learning objectives are only motivated empirically. Multi-Class Scoring Disagreement (MCSD) divergence is presented by aggregating the absolute margin violations in multi-class classification, and this proposed MCSD is able to fully characterize the relations between any pair of multi-class scoring hypotheses. By using MCSD as a measure of domain distance, we develop a new domain adaptation bound for multi-class UDA; its data-dependent, probably approximately correct bound is also developed that naturally suggests adversarial learning objectives to align conditional feature distributions across source and target domains. Consequently, an algorithmic framework of Multi-class Domain-adversarial learning Networks (McDalNets) is developed, and its different instantiations via surrogate learning objectives either coincide with or resemble a few recently popular methods, thus (partially) underscoring their practical effectiveness. Based on our identical theory for multi-class UDA, we also introduce a new algorithm of Domain-Symmetric Networks (SymmNets), which is featured by a novel adversarial strategy of domain confusion and discrimination. SymmNets affords simple extensions that work equally well under the problem settings of either closed set, partial, or open set UDA. We conduct careful empirical studies to compare different algorithms of McDalNets and our newly introduced SymmNets. Experiments verify our theoretical analysis and show the efficacy of our proposed SymmNets. In addition, we have made our implementation code publicly available. [ABSTRACT FROM AUTHOR]

We examine the problem of uncovering communities in complex real-world networks whose elements and their respective associations manifest as streams of data. Community detection is applied in emerging computational environments and concerns critical applications in diverse areas including social computing, web analysis, IoT and biology. Despite the already expended related research efforts, the task of revealing the community structure of massive and rapidly-evolving networks remains very challenging. More specifically, there is an emerging need for online approaches that ingest graph data as a stream. In this paper, we propose a streaming-graph community-detection algorithm that expands seed-sets of nodes to communities. We consider an online setting and process a stream of edges while aiming to form communities on-the-fly using partial knowledge of the graph structure. We use space-efficient structures to maintain very limited information regarding the nodes of the graph and the sought communities, so as to effectively process large scale networks. In addition to our novel streaming approach, we develop a technique that increases the accuracy of our algorithm considerably and additionally propose a new clustering algorithm that allows for automatically deriving the size of the communities we seek to detect. Using ground-truth communities for a wide range of large real-word and synthetic networks, our experimental evaluation shows that our approach does achieve accuracy comparable, and oftentimes better, to the state-of-the-art non-streaming community detection algorithms. More importantly, we attain significant improvements in both execution time and memory requirements. [ABSTRACT FROM AUTHOR]

Cooperative Vehicle-Infrastructure System (CVIS) can provide innovative services for traffic management and enable trips to be safer, more coordinated, and smarter. In the CVIS, the vehicles upload crowd sensing data to the Vehicular Edge Computing (VEC) server for quick data fusion and informed decision-making. However, with the ever-increasing number of vehicles, the VEC server cannot undertake massive computation-intensive tasks due to the limited edge computing capabilities. In this paper, we propose a reverse offloading framework that can fully utilize the vehicular computation resource to relieve the burden of the VEC server and further reduce the system latency. Under the proposed offloading framework, the binary reverse offloading (BRO) and partial reverse offloading (PRO) strategies are designed for two types of tasks, i.e., non-partitioned tasks and partitioned tasks. We formulate the system latency minimization problem by optimizing reverse offloading decisions, and the communication and computation resources allocation. Due to the non-convex and existing variables coupling, the original problem is transformed into the equivalent weighted-sum optimization problem. Based on the alternative optimization, we decouple the weighted-sum optimization problems into the two subproblems, and the closed-form expressions of transmission power and computation frequency of vehicles and RSU are derived. Low complexity greedy based efficient searching (GES) algorithm and joint alternative optimization based bi-section searching (JAOBSS) algorithm are proposed for BRO and PRO strategies, respectively. The algorithm complexity and performance bounds are analyzed. Simulation results show that the proposed GES algorithm can achieve optimal performance with low complexity. Besides, the proposed GES and JAOBSS algorithms can significantly improve the performance compared with other baseline schemes by 6.14% and 13.46% at least. [ABSTRACT FROM AUTHOR]

Dictionary learning has achieved remarkable success on a wide range of machine learning-based applications. In this paper, a locality-adaptive structured dictionary learning (LASDL) algorithm for cross-domain recognition is proposed. In the LASDL, a projective structured double reconstruction strategy is developed to train class-oriented sub-dictionaries from the specific classes of cross-domain samples. The strategy benefits to make full advantage of the discriminative information of cross-domain data and bridge the distribution divergence between two different domains. Meanwhile, an adaptive geometrical structure preserving function is designed to not only preserve the local manifold structures spanned by the representation coefficient spaces of the source and target domains, but also impose a constraint that the coefficients should keep closer to their class centers, which is propitious to reduce the distribution divergence and make the representation more accurate. With the structured linear coding technique, the final cross-domain recognition can be efficiently performed by determining class-specific reconstruction error. The optimization of the proposed LASDL model can be efficiently solved by simple least square method and alternating direction method of multipliers (ADMM) algorithm. Extensive experimental results validate the superiority of the proposed algorithm in contrast to other state-of-the-art predecessors. [ABSTRACT FROM AUTHOR]

Infrared and visible image fusion (IVIF) expects to obtain images that retain thermal radiation information from infrared images and texture details from visible images. In this paper, a model-based convolutional neural network (CNN) model, referred to as Algorithm Unrolling Image Fusion (AUIF), is proposed to overcome the shortcomings of traditional CNN-based IVIF models. The proposed AUIF model starts with the iterative formulas of two traditional optimization models, which are established to accomplish two-scale decomposition, i.e., separating low-frequency base information and high-frequency detail information from source images. Then the algorithm unrolling is implemented where each iteration is mapped to a CNN layer and each optimization model is transformed into a trainable neural network. Compared with the general network architectures, the proposed framework combines the model-based prior information and is designed more reasonably. After the unrolling operation, our model contains two decomposers (encoders) and an additional reconstructor (decoder). In the training phase, this network is trained to reconstruct the input image. While in the test phase, the base (or detail) decomposed feature maps of infrared/visible images are merged respectively by an extra fusion layer, and then the decoder outputs the fusion image. Qualitative and quantitative comparisons demonstrate the superiority of our model, which can robustly generate fusion images containing highlight targets and legible details, exceeding the state-of-the-art methods. Furthermore, our network has fewer weights and faster speed. [ABSTRACT FROM AUTHOR]

Community detection is one of the fundamental tasks in graph mining, which aims to identify group assignment of nodes in a complex network. Recently, network embedding techniques have demonstrated their strong power in advancing the community detection task and achieve better performance than various traditional methods. Despite their empirical success, most of the existing algorithms directly leverage the observed coarse network structure for community detection. Therefore, they often lead to suboptimal performance as the observed connections fail to capture the essential tie strength information among nodes precisely and account for the impact of noisy links. In this paper, an optimal network structure for community detection is introduced to characterize the fine-grained tie strength information between connected nodes and alleviate the adverse effects of noisy links. To obtain an expressive node representation for community detection, we learn the optimal network structure and network embeddings in a joint framework, instead of using a two-stage approach to derive the node embeddings from the coarse network topology. In particular, we formulate the joint framework as an optimization problem and an alternating optimization algorithm is exploited to solve the proposed optimization problem. Additionally, theoretical analyses regarding the computational complexity and the convergence of the optimization algorithm are also provided. Extensive experiments on both synthetic and real-world networks demonstrate the effectiveness and superiority of the proposed framework. [ABSTRACT FROM AUTHOR]

This paper studies the convergence properties of the well-known message-passing algorithm for convex optimisation. Under the assumption of pairwise separability and scaled diagonal dominance, asymptotic convergence is established and a simple bound for the convergence rate is provided for message-passing. In comparison with previous results, our results do not require the given convex program to have known convex pairwise components and that our bound for the convergence rate is tighter and simpler. When specialised to quadratic optimisation, we generalise known results by providing a very simple bound for the convergence rate. [ABSTRACT FROM AUTHOR]

In this paper, we consider the policy evaluation problem in reinforcement learning with agents on a decentralized and directed network. In order to evaluate the quality of a fixed policy in this decentralized setting, one option is for agents to run decentralized temporal-difference (TD) collaboratively. To account for the practical scenarios where the state and action spaces are large and malicious attacks emerge, we focus on the decentralized TD learning with linear function approximation in the presence of malicious agents (often termed as Byzantine agents). We propose a trimmed mean-based Byzantine-resilient decentralized TD algorithm to perform policy evaluation in this setting. We establish the finite-time convergence rate, as well as the asymptotic learning error that depends on the number of Byzantine agents. Numerical experiments corroborate the robustness of the proposed algorithm. [ABSTRACT FROM AUTHOR]

This paper presents algorithms for the parallelization of inference in hidden Markov models (HMMs). In particular, we propose a parallel forward-backward type of filtering and smoothing algorithm as well as a parallel Viterbi-type maximum-a-posteriori (MAP) algorithm. We define associative elements and operators to pose these inference problems as all-prefix-sums computations and parallelize them using the parallel-scan algorithm. The advantage of the proposed algorithms is that they are computationally efficient in HMM inference problems with long time horizons. We empirically compare the performance of the proposed methods to classical methods on a highly parallel graphics processing unit (GPU). [ABSTRACT FROM AUTHOR]

In this paper, we investigate designs of linear precoders for vector Gaussian channels via stochastic optimizations and deep neural networks (DNNs). We assume that the channel inputs are drawn from practical finite alphabets, and we search for precoders maximizing the mutual information between channel inputs and outputs. Though the problem is generally non-convex, we prove that when the right singular matrix of precoder is fixed, any local optima of this problem is a global optima. Based on this fact, an efficient projected stochastic gradient descent (PSGD) algorithm is designed to search the optimal precoders. Moreover, to reduce the complexity of calculating a posterior means involved in gradients calculation, K-best algorithm is adopted to make approximations of a posterior means with negligible loss of accuracy. Furthermore, to avoid explicit calculation of mutual information and its gradients, DNN-based autoencoders (AEs) are constructed for this precoding task, and an efficient training algorithm is proposed. We also prove that the AEs, with ‘softmax’ activation function and ‘categorical cross entropy’ loss, maximize the mutual information under reasonable assumptions. Then, in order to extend the AE methods to large scale systems, ‘sigmoid’ activation function and ‘binary cross entropy’ loss are used such that the size of AEs will not grow prohibitively large. We prove that this maximizes a lower bound of the mutual information under reasonable assumptions. Finally, to make the precoders practical for high speed wireless scenarios, we propose an offline training paradigm which trains DNNs to infer optimal precoders given channel state information instead of training online for every different channel. Simulation results show that all the proposed methods work well in maximizing mutual information and improving bit error rate (BER) performance. [ABSTRACT FROM AUTHOR]

In real-world recognition/classification tasks, limited by various objective factors, it is usually difficult to collect training samples to exhaust all classes when training a recognizer or classifier. A more realistic scenario is open set recognition (OSR), where incomplete knowledge of the world exists at training time, and unknown classes can be submitted to an algorithm during testing, requiring the classifiers to not only accurately classify the seen classes, but also effectively deal with unseen ones. This paper provides a comprehensive survey of existing open set recognition techniques covering various aspects ranging from related definitions, representations of models, datasets, evaluation criteria, and algorithm comparisons. Furthermore, we briefly analyze the relationships between OSR and its related tasks including zero-shot, one-shot (few-shot) recognition/learning techniques, classification with reject option, and so forth. Additionally, we also review the open world recognition which can be seen as a natural extension of OSR. Importantly, we highlight the limitations of existing approaches and point out some promising subsequent research directions in this field. [ABSTRACT FROM AUTHOR]

Black-box optimization algorithms typically start a search from scratch, assuming little prior knowledge about the task at hand. In practice, this approach can be prohibitive for computationally expensive problems, as a large number of costly function evaluations are often needed before a suitable (near-optimal) solution is found. Under this observation, recent efforts have incorporated transfer learning capabilities into sequential model-based Bayesian optimization (BO) solvers, resulting in substantial performance speed-ups by leveraging information from related past problems. However, a common simplifying assumption in existing approaches is that the search spaces of a previously encountered source and the ongoing target task bear the same features and dimensionality, with the difference lying in their respective objective functions. In this article, we present a generalized transfer BO algorithm that relaxes the aforementioned assumption. Our method jointly transforms source features while training probabilistic transfer regression models for the target, thus applying to practical use-cases where (in addition to the difference in objective functions) the number of features could change across the source and target tasks; for example, features can be added and/or removed. The theoretical basis of our proposal is analyzed, and its empirical performance is demonstrated on synthetic benchmark functions as well as in realistic examples spanning engineering design and the automated configuration of a machine learning model. Note to Practitioners—Problems of industrial interest have a tendency of being repetitive in nature. For this reason, domain experts are always in high demand, as they are able to harness their experience of similar problems to come up with fast solutions in difficult situations. However, domain experts are not easy to find. Given this fact, the present paper puts forth a method for automating the process of knowledge extraction (through experiential learning) and transfer across problems in the domain of computationally expensive black-box optimization. The key novelty and motivation of this work lies in enabling the adaptive transfer of knowledge even when the number of features changes across the source and target problems. Our proposed approach is verified experimentally on a range of benchmarks as well as real-world problems of a computationally expensive nature, highlighting the utility of an optimization engine that is able to learn from experience without the need for constant human intervention. [ABSTRACT FROM AUTHOR]

In this paper, we present a framework for performing automatic analysis of Land Use Zones based on Location-Based Social Networks (LBSNs). We model city areas using a hierarchical structure of POIs extracted from foursquare. We encode such structures in kernel machines, e.g., Support Vector Machines, using a new Tree Kernel, i.e., the Hierarchical POI Kernel (HPK), which can take the importance of the individual POIs into account during the substructure matching. This way, HPK projects structures in the space of all their possible substructures such that each dimension corresponds to a semantic structural feature, weighted according to the discriminative power of POIs. We generated four different datasets for the following cities: Barcelona, Lisbon, Amsterdam and Milan, where we trained and tested our models. The results show that our approach largely outperforms previous work and standard baseline built on simple features, such as counts of different POIs. Finally, we apply a mining algorithm to extract the most relevant features (tree fragments) from the implicit TK space according to the weights the kernel machine assigned to them. Our approach can produce an explicit set of representative features that can be used to classify and characterize urban areas. [ABSTRACT FROM AUTHOR]

With the explosive growth in the scale of cloud computing infrastructures, reliability and energy efficiency have become important concerns considering the great complexity of cloud data centers. There is an urgent need for efficient task assignment that can dispatch tasks to appropriate cloud data center servers, which is critical to achieve reliability and energy efficiency in current cloud data centers. Most of the research on task assignment focuses on only one of the objectives of reliability and energy efficiency, while the two objectives are intrinsically conflicting with each other. In this paper, we deal with the problem of task assignment in data centers, with the objective of minimizing the energy consumption while providing failure tolerance to task execution failure. We propose a reliability-aware and energy-efficient task replica assignment algorithm based on running task replicas at a low speed and enabling multiple task replicas to share the same server resources. Each task in a job processed by the cloud computing platform has two instances: main task and task replica (shadow). Each main task runs on an individual server, and the task replica associated with the main task is assigned on a different server. The main tasks run at the full server speed, while the task replicas run at a lower rate than the main tasks. The task replicas can be mapped onto dedicated backup servers or be assigned to the servers on which the main tasks are running. Multiple task replicas can share the same server resources to reduce the number of servers required. We conduct experiments through simulations. Experimental results demonstrate that the proposed algorithm can effectively reduce the energy consumption, while achieving a good balance between the number of servers used and job completion time. [ABSTRACT FROM AUTHOR]

Multi-label classification is an important research topic in machine learning, for which exploiting label dependencies is an effective modeling principle. Recently, probabilistic models have shown great potential in discovering dependencies among labels. In this paper, motivated by the recent success of multi-view learning to improve the generalization performance, we propose a novel multi-view probabilistic model named latent conditional Bernoulli mixture (LCBM) for multi-label classification. LCBM is a generative model taking features from different views as inputs, and conditional on the latent subspace shared by the views a Bernoulli mixture model is adopted to build label dependencies. Inside each component of the mixture, the labels have a weak correlation which facilitates computational convenience. The mean field variational inference framework is used to carry out approximate posterior inference in the probabilistic model, where we propose a Gaussian mixture variational autoencoder (GMVAE) for effective posterior approximation. We further develop a scalable stochastic training algorithm for efficiently optimizing the model parameters and variational parameters, and derive an efficient prediction procedure based on greedy search. Experimental results on multiple benchmark datasets show that our approach outperforms other state-of-the-art methods under various metrics. [ABSTRACT FROM AUTHOR]

Incomplete multi-view clustering (IMVC) optimally combines multiple pre-specified incomplete views to improve clustering performance. Among various excellent solutions, the recently proposed multiple kernel $k$ k -means with incomplete kernels (MKKM-IK) forms a benchmark, which redefines IMVC as a joint optimization problem where the clustering and kernel matrix imputation tasks are alternately performed until convergence. Though demonstrating promising performance in various applications, we observe that the manner of kernel matrix imputation in MKKM-IK would incur intensive computational and storage complexities, over-complicated optimization and limitedly improved clustering performance. In this paper, we first propose an Efficient and Effective Incomplete Multi-view Clustering (EE-IMVC) algorithm to address these issues. Instead of completing the incomplete kernel matrices, EE-IMVC proposes to impute each incomplete base matrix generated by incomplete views with a learned consensus clustering matrix. Moreover, we further improve this algorithm by incorporating prior knowledge to regularize the learned consensus clustering matrix. Two three-step iterative algorithms are carefully developed to solve the resultant optimization problems with linear computational complexity, and their convergence is theoretically proven. After that, we theoretically study the generalization bound of the proposed algorithms. Furthermore, we conduct comprehensive experiments to study the proposed algorithms in terms of clustering accuracy, evolution of the learned consensus clustering matrix and the convergence. As indicated, our algorithms deliver their effectiveness by significantly and consistently outperforming some state-of-the-art ones. [ABSTRACT FROM AUTHOR]

Drones have a wide range of applications in urban environments as they can both enhance people's daily activities and commercial activities through various operations and deployments. With the increasing number of drones, flight safety and efficiency become the main concern, and effective drone operations can make a difference. Accordingly, 4D path planning for drone operations is the focus of this paper, and the swarm-based method is proposed to solve this complicated optimization problem. Under the framework of ‘AirMatrix’, the problem is solved in two levels, i.e., 3D path planning for a single drone and conflict resolution among drones. In the multi-path planning level, multiple alternative flight paths for each drone are generated to increase the acceptance rate of a flight request. The constraints on a single flight path and two different flight paths are considered. The goal is to obtain several different short flight paths as alternatives. A clustering improved ant colony optimization CIACO) algorithm is employed to solve the multi-path planning problem. The crowding mechanism is used in clustering, and some improvements are made to strengthen the global and local search ability in the early and later phases of iterations. In the task scheduling level, the conflicts between two drones are defined in two circumstances. One is for the time interval of passing the same path point, another one is for the right-angle collision between two drones. A three-layer fitness function is proposed to maximize the number of permitted flights according to the safety requirement, in which the airspace utilization and the operators’ requests are both considered. A ‘cross-off’ strategy is developed to calculate the fitness value, and a ‘distributed-centralized’ strategy is applied considering the task priorities of drones. A genetic algorithm GA)-based task scheduling algorithm is also developed according to the characteristic of the established model. Simulation results demonstrate that 4D flight path of each drone can be generated by the proposed swarmed-based algorithms, and safe and efficient drone operations in a specific airspace can be ensured. [ABSTRACT FROM AUTHOR]

To meet the execution requirements of delay-sensitive services in vehicular edge computing (VEC) networks, vehicular services need to be offloaded to edge computing nodes. For complex, large-scale services, the services need to be migrated if the services are not completed before the vehicles leave the coverage of edge computing nodes. Trust and resource matching between areas thus become major problems. This paper studies the decision model of vehicular service offloading and migration. First, software-defined network (SDN) technology is introduced into the traditional network architecture, and a two-layer distributed SDN-controlled VEC network architecture is designed, which is divided into a domain control layer and an area control layer. In this framework, we use the consortium blockchain as a carrier to share network topology information between SDN controllers to prevent information leakage. We then established a service offloading and migration optimization problem model to minimize service execution delay, reduce energy consumption and maximize the throughput of the blockchain system. We describe the problem model as a Markov Decision Process (MDP), introduce a deep reinforcement learning (DRL) algorithm named asynchronous advantage actor-critic (A3C) and design a dynamic service offloading and migration algorithm (DSOMA) based on A3C to solve the problem. Simulation results show that DSOMA can increase the throughput of the blockchain system, and DSOMA is superior to the deep Q-learning (DQN) algorithm and greedy offloading algorithm in reducing service execution delay and system energy consumption. [ABSTRACT FROM AUTHOR]

Articulation angle sensing is an essential component of manoeuvrability and stability control systems for articulated heavy goods vehicles, particularly long combination vehicles. Existing solutions to this sensing task are limited by reliance on trailer modifications or information or by measurement accuracy, or both, restricting commercial adoption. In this paper we present a purely tractor-based sensor concept comprising a rear-facing camera and the parallel tracking and mapping (PTAM) image processing algorithm. The system requires no prior knowledge of or modifications to the trailer, is compatible with planar and non-planar trailer shapes, and with multiply-articulated vehicle combinations. The system is validated in full-scale vehicle tests on both a tractor semi-trailer combination and a truck and full-trailer combination, demonstrating robust performance in a number of conditions, including trailers with non-planar geometry and with minimal visual features. Average RMS measurement errors of 1.19, 1.03 and 1.53 degrees were demonstrated for the semi-trailer and full-trailer (drawbar and semi-trailer) respectively. This compares favourably with the state-of-the-art in the published literature. A number of improvements are proposed for future development based on the observations in this research. [ABSTRACT FROM AUTHOR]

In recent years, unmanned aerial vehicles (UAVs) have unlocked numerous sensing applications, which are expected to add billions of dollars to the world economy in the next decade. To further improve the Quality-of-Service in these applications, the 3rd Generation Partnership Project has considered the use of terrestrial cellular networks to support UAV sensing services, also known as the cellular Internet of UAVs. In this paper, we consider a cellular Internet of UAVs, where the sensory data can be transmitted either to the base station via cellular links, or to the mobile devices by underlay UAV-to-Device (U2D) communications. To evaluate the freshness of the sensory data, the concept of age of information (AoI) is adopted, in which a lower AoI implies fresher data. Since UAVs’ AoIs are determined by their trajectories during sensing and transmission, we investigate the AoI minimization problem for UAVs by designing their trajectories. This problem is a Markov decision problem with an infinite state-action space, and thus we utilize multi-agent deep reinforcement learning to approximate the state-action space. Then, we propose a multi-UAV trajectory design algorithm to solve this problem. Simulation results show that our proposed algorithm can achieve a lower AoI than a greedy algorithm, policy gradient algorithm, and overlay U2D scheme. [ABSTRACT FROM AUTHOR]

Deep neural networks (DNNs) have been widely applied for classification tasks in wireless communications such as link scheduling, user association and base station (BS) muting. When training a DNN using datasets with highly-skewed class distribution where most data belong to a few majority classes, learning performance for minority classes will degrade since the imbalanced data forces the training process to be biased towards the majority classes. In many wireless problems, multiple decisions for classification are made jointly, say deciding which BSs in a network should be muted to avoid interference. In this paper, we employ DNN to learn an optimal predictive interference coordination policy, and strive to avoid biased learning in such a multi-label multi-class classification problem. The major contribution is proposing a training method to encourage fairness among classes by minimizing the maximal cost of decisions among classes, which is converted into a problem to optimize the weighting factors on the training cost of each class. We provide a gradient descent algorithm to optimize the weighting factors and the model parameters of the DNN by alternative updates. Simulation results show that the training method can improve the learning performance for the minority class and can achieve higher network utility than existing training methods. [ABSTRACT FROM AUTHOR]

Defocus blur detection (DBD), which has been widely applied to various fields, aims to detect the out-of-focus or in-focus pixels from a single image. Despite the fact that the deep learning based methods applied to DBD have outperformed the hand-crafted feature based methods, the performance cannot still meet our requirement. In this paper, a novel network is established for DBD. Unlike existing methods which only learn the projection from the in-focus part to the ground-truth, both in-focus and out-of-focus pixels, which are completely and symmetrically complementary, are taken into account. Specifically, two symmetric branches are designed to jointly estimate the probability of focus and defocus pixels, respectively. Due to their complementary constraint, each layer in a branch is affected by an attention obtained from another branch, effectively learning the detailed information which may be ignored in one branch. The feature maps from these two branches are then passed through a unique fusion block to simultaneously get the two-channel output measured by a complementary loss. Additionally, instead of estimating only one binary map from a specific layer, each layer is encouraged to estimate the ground truth to guide the binary map estimation in its linked shallower layer followed by a top-to-bottom combination strategy, gradually exploiting the global and local information. Experimental results on released datasets demonstrate that our proposed method remarkably outperforms state-of-the-art algorithms. [ABSTRACT FROM AUTHOR]

Disparity estimation is an essential task taking part in many light-field applications. Due to the complexity of algorithms and high dimensional property of light-field data, performing this task involves a significant computational effort and results in very long processing time on CPU. Graphics processing units (GPUs), which is capable of massively parallel processing, is a promising solution to cover the computation requirement and speed up the task. In this paper, we develop a GPU-accelerated approach for light-field disparity estimation using a variational computation framework (GVLD). Our algorithm combines the intrinsic sub-pixel precision of variational formulation and the effectiveness of weighted median filtering to produce a highly accurate solution. The proposed algorithm is fully parallelized and optimized for the implementation using the OpenCL framework. An intensive evaluation including a quantitative comparison to related works and a detailed analysis of the proposed approach’s performance is presented. Experimental results demonstrate our superior performance compared to state-of-the-art approaches. The proposed approach is 10+ times faster than other approaches running on a similar GPU platform and provides the most accurate solution among optimization-based approaches. Compared to the implementation running on CPU, our GPU-accelerated method achieves up to $365\times $ speed up. [ABSTRACT FROM AUTHOR]

Hyperspectral image (HSI) clustering is a challenging task due to the complex characteristics in HSI data, such as spatial-spectral structure, high-dimension, and large spectral variability. In this paper, we propose a novel deep spatial-spectral subspace clustering network (DS3C-Net), which explores spatial-spectral information via the multi-scale auto-encoder and collaborative constraint. Considering the structure correlations of HSI, the multi-scale auto-encoder is first designed to extract spatial-spectral features with different-scale pixel blocks which are selected as the inputs. Then, the collaborative constrained self-expressive layers are introduced between the encoder and decoder, to capture the self-expressive subspace structures. By designing a self-expressiveness similarity constraint, the proposed network is trained collaboratively, and the affinity matrices of the feature representation are learned in an end-to-end manner. Based on the affinity matrices, the spectral clustering algorithm is utilized to obtain the final HSI clustering result. Experimental results on three widely used hyperspectral image datasets demonstrate that the proposed method outperforms state-of-the-art methods. [ABSTRACT FROM AUTHOR]

As the training dataset size and the model size of machine learning increase rapidly, more computing resources are consumed to speedup the training process. However, the scalability and performance reproducibility of parallel machine learning training, which mainly uses stochastic optimization algorithms, are limited. In this paper, we demonstrate that the sample difference in the dataset plays a prominent role in the scalability of parallel machine learning algorithms. We propose to use statistical properties of dataset to measure sample differences. These properties include the variance of sample features, sample sparsity, sample diversity, and similarity in sampling sequences. We choose four types of parallel training algorithms as our research objects: (1) the asynchronous parallel SGD algorithm (Hogwild! algorithm), (2) the parallel model average SGD algorithm (minibatch SGD algorithm), (3) the decentralization optimization algorithm, and (4) the dual coordinate optimization (DADM algorithm). Our results show that the statistical properties of training datasets determine the scalability upper bound of these parallel training algorithms. [ABSTRACT FROM AUTHOR]