The minimum independent dominance set (MIDS) problem is an important version of the dominating set with some other applications. In this work, we present an improved master-apprentice evolutionary algorithm for solving the MIDS problem based on a path-breaking strategy called MAE-PB. The proposed MAE-PB algorithm combines a construction function for the initial solution generation and candidate solution restarting. It is a multiple neighborhood-based local search algorithm that improves the quality of the solution using a path-breaking strategy for solution recombination based on master and apprentice solutions and a perturbation strategy for disturbing the solution when the algorithm cannot improve the solution quality within a certain number of steps. We show the competitiveness of the MAE-PB algorithm by presenting the computational results on classical benchmarks from the literature and a suite of massive graphs from real-world applications. The results show that the MAE-PB algorithm achieves high performance. In particular, for the classical benchmarks, the MAE-PB algorithm obtains the best-known results for seven instances, whereas for several massive graphs, it improves the best-known results for 62 instances. We investigate the proposed key ingredients to determine their impact on the performance of the proposed algorithm.

A virtual cosmetics try-on system provides a realistic try-on experience for consumers and helps them efficiently choose suitable cosmetics. In this article, we propose a real-time augmented reality virtual cosmetics try-on system for smartphones (ARCosmetics), taking speed, accuracy, and stability into consideration at each step to ensure a better user experience. A novel and very fast face tracking method utilizes the face detection box and the average position of facial landmarks to estimate the faces in continuous frames. A dynamic weight Wing loss is introduced to assign a dynamic weight to every landmark by the estimated error during training. It balances the attention between small, medium, and large range error and thus increases the accuracy and robustness. We also designed a weighted average method to utilize the information of the adjacent frame for landmark refinement, guaranteeing the stability of the generated landmarks. Extensive experiments conducted on a large 106-point facial landmark dataset and the 300-VW dataset demonstrate the superior performance of the proposed method compared to other state-of-the-art methods. We also conducted user satisfaction studies further to verify the efficiency and effectiveness of our ARCosmetics system.

Although matrix multiplication plays an essential role in a wide range of applications, previous works only focus on optimizing dense or sparse matrix multiplications. The Sparse Approximate Matrix Multiply (SpAMM) is an algorithm to accelerate the multiplication of decay matrices, the sparsity of which is between dense and sparse matrices. In addition, large-scale decay matrix multiplication is performed in scientific applications to solve cutting-edge problems. To optimize large-scale decay matrix multiplication using SpAMM on supercomputers such as Sunway Taihulight, we present swSpAMM, an optimized SpAMM algorithm by adapting the computation characteristics to the architecture features of Sunway Taihulight.

Specifically, we propose both intra-node and inter-node optimizations to accelerate swSpAMM for large-scale execution. For intra-node optimizations, we explore algorithm parallelization and block-major data layout that are tailored to better utilize the architecture advantage of Sunway processor. For inter-node optimizations, we propose a matrix organization strategy for better distributing sub-matrices across nodes and a dynamic scheduling strategy for improving load balance across nodes. We compare swSpAMM with the existing GEMM library on a single node as well as large-scale matrix multiplication methods on multiple nodes. The experiment results show that swSpAMM achieves a speedup up to 14.5× and 2.2× when compared to xMath library on a single node and 2D GEMM method on multiple nodes, respectively.

With the scaling up of high-performance computing systems in recent years, their reliability has been descending continuously. Therefore, system resilience has been regarded as one of the critical challenges for large-scale HPC systems. Various techniques and systems have been proposed to ensure the correct execution and completion of parallel programs. This paper provides a comprehensive survey of existing software resilience approaches. Firstly, a classification of software resilience approaches is presented; then we introduce major approaches and techniques, including checkpointing, replication, soft error resilience, algorithm-based fault tolerance, fault detection and prediction. In addition, challenges exposed by system-scale and heterogeneous architecture are also discussed.

Meta-learning has been widely applied to solving few-shot reinforcement learning problems, where we hope to obtain an agent that can learn quickly in a new task. However, these algorithms often ignore some isolated tasks in pursuit of the average performance, which may result in negative adaptation in these isolated tasks, and they usually need sufficient learning in a stationary task distribution. In this paper, our algorithm presents a hierarchical framework of double meta-learning, and the whole framework includes classification, meta-learning, and re-adaptation. Firstly, in the classification process, we classify tasks into several task subsets, considered as some categories of tasks, by learned parameters of each task, which can separate out some isolated tasks thereafter. Secondly, in the meta-learning process, we learn category parameters in all subsets via meta-learning. Simultaneously, based on the gradient of each category parameter in each subset, we use meta-learning again to learn a new meta-parameter related to the whole task set, which can be used as an initial parameter for the new task. Finally, in the re-adaption process, we adapt the parameter of the new task with two steps, by the meta-parameter and the appropriate category parameter successively. Experimentally, we demonstrate our algorithm prevents the agent from negative adaptation without losing the average performance for the whole task set. Additionally, our algorithm presents a more rapid adaptation process within re-adaptation. Moreover, we show the good performance of our algorithm with fewer samples as the agent is exposed to an online meta-learning setting.

Label distribution learning (LDL) is a new learning paradigm to deal with label ambiguity and many researches have achieved the prominent performances. Compared with traditional supervised learning scenarios, the annotation with label distribution is more expensive. Direct use of existing active learning (AL) approaches, which aim to reduce the annotation cost in traditional learning, may lead to the degradation of their performance. To deal with the problem of high annotation cost in LDL, we propose the active label distribution learning via kernel maximum mean discrepancy (ALDL-kMMD) method to tackle this crucial but rarely studied problem. ALDL-kMMD captures the structural information of both data and label, extracts the most representative instances from the unlabeled ones by incorporating the nonlinear model and marginal probability distribution matching. Besides, it is also able to markedly decrease the amount of queried unlabeled instances. Meanwhile, an effective solution is proposed for the original optimization problem of ALDL-kMMD by constructing auxiliary variables. The effectiveness of our method is validated with experiments on the real-world datasets.

Innovations on the Internet of Everything (IoE) enabled systems are driving a change in the settings where we interact in smart units, recognized globally as smart city environments. However, intelligent video-surveillance systems are critical to increasing the security of these smart cities. More precisely, in today’s world of smart video surveillance, person re-identification (Re-ID) has gained increased consideration by researchers. Various researchers have designed deep learning-based algorithms for person Re-ID because they have achieved substantial breakthroughs in computer vision problems. In this line of research, we designed an adaptive feature refinement-based deep learning architecture to conduct person Re-ID. In the proposed architecture, the inter-channel and inter-spatial relationship of features between the images of the same individual taken from nonidentical camera viewpoints are focused on learning spatial and channel attention. In addition, the spatial pyramid pooling layer is inserted to extract the multiscale and fixed-dimension feature vectors irrespective of the size of the feature maps. Furthermore, the model’s effectiveness is validated on the CUHK01 and CUHK02 datasets. When compared with existing approaches, the approach presented in this paper achieves encouraging Rank 1 and 5 scores of 24.6% and 54.8%, respectively.

In the minimum degree vertex deletion problem, we are given a graph, a distinguished vertex in the graph, and an integer \kappa, and the question is whether we can delete at most \kappa vertices from the graph so that the distinguished vertex has the unique minimum degree. The maximum degree vertex deletion problem is defined analogously but here we want the distinguished vertex to have the unique maximum degree. It is known that both problems are \mathsf{\text{NP}}-hard and fixed-parameter intractable with respect to some natural parameters. In this paper, we study the (parameterized) complexity of these two problems restricted to split graphs, p-degenerate graphs, and planar graphs. Our study provides a comprehensive complexity landscape of the two problems restricted to these special graphs.

Mobile opportunistic network (MON) is an efficient way of communication when there is no persistent connection between nodes. Multicast in MONs can be used to efficiently deliver messages to multiple destination nodes. However, because multiple destination nodes are involved, multicast routing is more complex than unicast and brings a higher communication cost. Backbone-based routing can effectively reduce the network overhead and the complexity of routing scheme. However, the load of backbone nodes is larger than that of regular nodes. If the backbone node’s buffer is exhausted, it will have a significant impact on the performance of the routing scheme. Load balancing can improve the ability of backbone to deal with the change of network load, and backbone maintenance algorithm can provide backbone robustness. In this paper, we propose a robust load-balanced backbone-based multicast routing scheme in MONs. In the backbone construction algorithm, we transform the problem of backbone construction into a multi-objective optimization problem, and propose a multi-objective evolutionary algorithm-based backbone construction algorithm, namely LBMBC-MOEA algorithm. In addition, in order to increase the robustness of the backbone-based routing scheme, we propose a localized multicast backbone maintenance algorithm (MBMA) to deal with the buffer exhaustion of backbone nodes. When a backbone node’s residual buffer is insufficient, MBMA algorithm selects other nodes to replace the backbone node. The results on extensive simulations show that when considering the node buffer size constraints, compared with previous backbone-based multicast routing schemes, our proposed algorithm has better performance, and when the node’s residual buffer is insufficient, MBMA algorithm can significantly improve the performance of the backbone-based multicast routing scheme.

With the rapid development of mobile technology and smart devices, crowdsensing has shown its large potential to collect massive data. Considering the limitation of calculation power, edge computing is introduced to release unnecessary data transmission. In edge-computing-enabled crowdsensing, massive data is required to be preliminary processed by edge computing devices (ECDs). Compared with the traditional central platform, these ECDs are limited by their own capability so they may only obtain part of relative factors and they can’t process data synthetically. ECDs involved in one task are required to cooperate to process the task data. The privacy of participants is important in crowdsensing, so blockchain is used due to its decentralization and tamper-resistance. In crowdsensing tasks, it is usually difficult to obtain the assessment criteria in advance so reinforcement learning is introduced. As mentioned before, ECDs can’t process task data comprehensively and they are required to cooperate quality assessment. Therefore, a blockchain-based framework for data quality in edge-computing-enabled crowdsensing (BFEC) is proposed in this paper. DPoR (Delegated Proof of Reputation), which is proposed in our previous work, is improved to be suitable in BFEC. Iteratively, the final result is calculated without revealing the privacy of participants. Experiments on the open datasets Adult, Blog, and Wine Quality show that our new framework outperforms existing methods in executing sensing tasks.

In analytical queries, a number of important operators like JOIN and GROUP BY are suitable for parallelization, and GPU is an ideal accelerator considering its power of parallel computing. However, when data size increases to hundreds of gigabytes, one GPU card becomes insufficient due to the small capacity of global memory and the slow data transfer between host and device. A straightforward solution is to equip more GPUs linked with high-bandwidth connectors, but the cost will be highly increased. We utilize unified memory (UM) produced by NVIDIA CUDA (Compute Unified Device Architecture) to make it possible to accelerate large-scale queries on just one GPU, but we notice that the transfer performance between host and UM, which happens before kernel execution, is often significantly slower than the theoretical bandwidth. An important reason is that, in single-GPU environment, data processing systems usually invoke only one or a static number of threads for data copy, leading to an inefficient transfer which slows down the overall performance heavily. In this paper, we present D-Cubicle, a runtime module to accelerate data transfer between host-managed memory and unified memory. D-Cubicle boosts the actual transfer speed dynamically through a self-adaptive approach. In our experiments, taking data transfer into account, D-Cubicle processes 200 GB of data on a single GPU with 32 GB of global memory, achieving 1.43x averagely and 2.09x maximally the performance of the baseline system.

Rumor detection has become an emerging and active research field in recent years. At the core is to model the rumor characteristics inherent in rich information, such as propagation patterns in social network and semantic patterns in post content, and differentiate them from the truth. However, existing works on rumor detection fall short in modeling heterogeneous information, either using one single information source only (e.g., social network, or post content) or ignoring the relations among multiple sources (e.g., fusing social and content features via simple concatenation).

Therefore, they possibly have drawbacks in comprehensively understanding the rumors, and detecting them accurately. In this work, we explore contrastive self-supervised learning on heterogeneous information sources, so as to reveal their relations and characterize rumors better. Technically, we supplement the main supervised task of detection with an auxiliary self-supervised task, which enriches post representations via post self-discrimination.

Specifically, given two heterogeneous views of a post (i.e., representations encoding social patterns and semantic patterns), the discrimination is done by maximizing the mutual information between different views of the same post compared to that of other posts. We devise cluster-wise and instance-wise approaches to generate the views and conduct the discrimination, considering different relations of information sources. We term this framework as self-supervised rumor detection (SRD). Extensive experiments on three real-world datasets validate the effectiveness of SRD for automatic rumor detection on social media.

Multi-source domain adaptation utilizes multiple source domains to learn the knowledge and transfers it to an unlabeled target domain. To address the problem, most of the existing methods aim to minimize the domain shift by auxiliary distribution alignment objectives, which reduces the effect of domain-specific features. However, without explicitly modeling the domain-specific features, it is not easy to guarantee that the domain-invariant representation extracted from input domains contains domain-specific information as few as possible. In this work, we present a different perspective on MSDA, which employs the idea of feature elimination to reduce the influence of domain-specific features. We design two different ways to extract domain-specific features and total features and construct the domain-invariant representations by eliminating the domain-specific features from total features. The experimental results on different domain adaptation datasets demonstrate the effectiveness of our method and the generalization ability of our model.

The construction of the tweakable Even-Mansour cipher is in fact the designs of permutations, mask operations, and masking functions. For information-theoretic security, permutations are usually taken as random permutations. This paper focuses on the mask operations and masking functions to construct a universal tweakable Even-Mansour cipher. Firstly, we describe a formal definition of a universal masking function and provide a universal tweakable Even-Mansour cipher UTEM. In the random permutation model, we prove that UTEM is multi-key secure by H-coefficients technique. Then we show some efficient instantiations of the universal masking function to concertize UTEM. Finally, we apply UTEM to an encryption mode TIE (tweak incrementation encryption) and an authenticated encryption mode IAPM (integrity aware parallelizable mode), present two new schemes TIE-plus and IAPM-plus, and prove their security. UTEM enriches tweakable blockciphers, brings more research topics, and plays an important role in modes of operation, which will be of great significance.

At ToSC 2019, Ankele et al. proposed a novel idea for constructing zero-correlation linear distinguishers in a related-tweakey model. This paper further clarifies this principle and gives a search model for zero-correlation distinguishers. As a result, for the first time, the authors construct 14-round and 16-round zero-correlation linear distinguishers for SKINNY-n-2n and SKINNY-n-3n, respectively, which are both two rounds longer than Anekele et al.’s. Based on these distinguishers, the paper presents related-tweakey zero-correlation linear attacks on 21-round SKINNY-n-2n and 25-round SKINNY-n-3n, respectively.

Differential privacy has recently become a widely recognized strict privacy protection model of data release. Differential privacy histogram publishing can directly show the statistical data distribution under the premise of ensuring user privacy for data query, sharing, and analysis. The dynamic data release is a study with a wide range of current industry needs. However, the amount of data varies considerably over different periods. Unreasonable data processing will result in the risk of users’ information leakage and unavailability of the data. Therefore, we designed a differential privacy histogram publishing method based on the dynamic sliding window of LSTM (DPHP-DL), which can improve data availability on the premise of guaranteeing data privacy. DPHP-DL is integrated by DSW-LSTM and DPHK+. DSW-LSTM updates the size of sliding windows based on data value prediction via long short-term memory (LSTM) networks, which evenly divides the data stream into several windows. DPHK+ heuristically publishes non-isometric histograms based on k-mean++ clustering of automatically obtaining the optimal K, so as to achieve differential privacy histogram publishing of dynamic data. Extensive experiments on real-world dynamic datasets demonstrate the superior performance of the DPHP-DL.