Recent studies have highlighted audio adversarial examples as a ubiquitous threat to state-of-the-art automatic speech recognition systems. Thorough studies on how to effectively generate adversarial examples are essential to prevent potential attacks. Despite many research on this, the efficiency and the robustness of existing works are not yet satisfactory. In this paper, we propose weighted-sampling audio adversarial examples, focusing on the numbers and the weights of distortion to reinforce the attack. Further, we apply a denoising method in the loss function to make the adversarial attack more imperceptible. Experiments show that our method is the first in the field to generate audio adversarial examples with low noise and high audio robustness at the minute time-consuming level 1.

Vehicular edge computing (VEC) is a new computing paradigm that has great potential to enhance the capability of vehicle terminals (VTs) to support resource-hungry in-vehicle applications with low latency and high energy efficiency. In this article, we investigate an important computation offloading scheduling problem in a typical VEC scenario, where a VT traveling along an expressway intends to schedule its tasks waiting in the queue to minimize the long-term cost in terms of a tradeoff between task latency and energy consumption. Due to diverse task characteristics, dynamic wireless environment, and frequent handover events caused by vehicle movements, an optimal solution should take into account both where to schedule (i.e., local computation or offloading) and when to schedule (i.e., the order and time for execution) each task. To solve such a complicated stochastic optimization problem, we model it by a carefully designed Markov decision process (MDP) and resort to deep reinforcement learning (DRL) to deal with the enormous state space. Our DRL implementation is designed based on the state-of-the-art proximal policy optimization (PPO) algorithm. A parameter-shared network architecture combined with a convolutional neural network (CNN) is utilized to approximate both policy and value function, which can effectively extract representative features. A series of adjustments to the state and reward representations are taken to further improve the training efficiency. Extensive simulation experiments and comprehensive comparisons with six known baseline algorithms and their heuristic combinations clearly demonstrate the advantages of the proposed DRL-based offloading scheduling method.

In nowadays, as the development of digital photographic technology, video files grow rapidly, there is a great demand for automatic video semantic analysis in many scenes, such as video semantic understanding, content-based analysis, video retrieval. Shot boundary detection is a key basic technology and first step for video analysis. However, recent methods are time consuming and performs bad in the gradual transition detection. In this paper we proposed a new approach which used CNN model to extract features of video sequence parallelly based on GPU, so we can simplify the expression of video and reduce the calculation time for shot detection, and took local frame similarity and dual-threshold sliding window similarity into consideration to increase recall and precise of shot detection. The experimental result shows that the proposed method can achieve a high F1 score and excellent detection speed.

Rough sets are efficient to extract rules from information systems. Matroids generalize the linear independency in vector spaces and the cycle in graphs. Specifically, matroids provide well-established platforms for greedy algorithms, while most existing algorithms for many rough set problems including attribute reduction are greedy ones. Therefore, the combination between rough sets and matroids may bring new efficient solutions to those important and difficult problems. In this paper, 2-circuit matroids, abstracted from matroidal characteristics of rough sets, are studied and axiomatized. A matroid is induced by an equivalence relation, and its characteristics including the independent set and duality are represented with rough sets. Based on these rough set representations, this special type of matroid is defined as 2-circuit matroids. Conversely, an equivalence relation is induced by a matroid, and its relationship with the above induction is further investigated. Finally, a number of axioms of the 2-circuit matroid are obtained through rough sets. These interesting and diverse axioms demonstrate the potential for the connection between rough sets and matroids.

Node localisation is a fundamental problem in wireless sensor networks. Many applications require the location information of sensor nodes. Received signal strength (RSS) is a simple and inexpensive approach for localisation purpose. However, the accuracy of RSS measurement is unpredictable owing to the nature of the radio frequency (RF) channel. An RSS-based Monte Carlo localisation scheme is proposed to sequentially estimate the location of mobile nodes, using the log-normal statistical model of RSS measurement. The RSS measurement is treated as the observation model in Monte Carlo method and the mobility feature of nodes as the transition model. Our method is widely applicable because the RSS function is easy to implement on nodes, and the mathematical model for mobile nodes may have non-analytic forms. Simulation results about localisation accuracy and cost show that this scheme is better than other methods.

Brain-computer interface (BCI) is a communication system that connects the brain with the computer and the peripheral equipment. In classification experiment of single-trial electroencephalogram (EEG) for left and right finger movement task, common spatial patterns (CSP) are employed to extract feature for EEG signals, and support vector machines (SVM) are used to classify. Basing on neurophysiological background of EEG signals, a new feature extraction method is proposed to select channel number, position, filter frequency and spatial filter number. Basing on analyzing change feature of the error penalty parameterCand the Gaussian kernel parameter sigma on support vector machines, a new area search table is proposed to improve classification accuracy.

Computer viruses can disable computer systems not only by destroying data or modifying a system's configuration, but also by consuming most of the computing resources such as CPU time and storage. The latter effects are related to the computational complexity of computer viruses. In this correspondence, we investigate some issues concerning the time complexity of computer viruses, and prove some known experimental results mathematically. We prove that there exist computer viruses with arbitrarily long running time, not only in the infecting procedure but in the executing procedure. Moreover, we prove that there are computer viruses with arbitrarily large time complexity in the detecting procedure, and there are undecidable computer viruses that have no "minimal" detecting procedure.