Current Issue : July - September Volume : 2019 Issue Number : 3 Articles : 5 Articles
Many IoT (Internet of Things) systems run Android systems or Android-like systems.\nWith the continuous development of machine learning algorithms, the learning-based Android\nmalware detection system for IoT devices has gradually increased. However, these learning-based\ndetection models are often vulnerable to adversarial samples. An automated testing framework is\nneeded to help these learning-based malware detection systems for IoT devices perform security\nanalysis. The current methods of generating adversarial samples mostly require training parameters\nof models and most of the methods are aimed at image data. To solve this problem, we propose a\ntesting framework for learning-based Android malware detection systems (TLAMD) for IoT Devices.\nThe key challenge is how to construct a suitable fitness function to generate an effective adversarial\nsample without affecting the features of the application. By introducing genetic algorithms and some\ntechnical improvements, our test framework can generate adversarial samples for the IoT Android\napplication with a success rate of nearly 100% and can perform black-box testing on the system....
Program slicing is a technique to extract the part of a program (the slice) that influences or is influenced by a set of variables at a\ngiven point (the slicing criterion). Computing minimal slices is undecidable in the general case, and obtaining the minimal slice\nof a given program is normally computationally prohibitive even for very small programs. Therefore, no matter what program\nslicer we use, in general, we cannot be sure that our slices are minimal. This is probably the fundamental reason why no\nbenchmark collection of minimal program slices exists. In this work, we present a method to automatically produce quasiminimal\nslices. Using our method, we have produced a suite of quasi-minimal slices for Erlang that we have later manually\nproved they are minimal. We explain the process of constructing the suite, the methodology and tools that were used, and the\nresults obtained. The suite comes with a collection of Erlang benchmarks together with different slicing criteria and the\nassociated minimal slices....
Big data and its analysis have become a widespread practice in recent times, applicable to multiple industries. Data mining is a\ntechnique that is based on statistical applications. This method extracts previously undetermined data items from large quantities\nof data. The banking and insurance industries use data mining analysis to detect fraud, offer the appropriate credit or insurance\nsolutions to customers, and better understand customer demands. This study aims to identify data mining classification algorithms\nand use them to predict default risks, avoid possible payment difficulties, and reduce potential problems in extending\ncredit. The data for this study, which contains demographic and socioeconomic characteristics of individuals, were obtained from\nthe Turkish Statistical Institute 2015 survey. Six classification algorithmsâ??Naive Bayes, Bayesian networks, J48, random forest,\nmultilayer perceptron, and logistic regressionâ??were applied to the dataset using WEKA 3.9 data mining software. These algorithms\nwere compared considering the root mean error squares, receiver operating characteristic area, accuracy, precision,\nF-measure, and recall statistical criteria. The best algorithmâ??logistic regressionâ??was obtained and applied to the real dataset to\ndetermine the attributes causing the default risk by using odds ratios. The socioeconomic and demographic characteristics of the\nindividuals were examined, and based on the odds ratio values, the results of which individuals and characteristics were more\nlikely to default, were reached. These results are not only beneficial to the literature but also have a significant influence in the\nfinancial industry in terms of the ability to predict customersâ?? default risk....
Deep learning solutions are being increasingly used in mobile applications. Although\nthere are many open-source software tools for the development of deep learning solutions, there are\nno guidelines in one place in a unified manner for using these tools toward real-time deployment\nof these solutions on smartphones. From the variety of available deep learning tools, the most\nsuited ones are used in this paper to enable real-time deployment of deep learning inference\nnetworks on smartphones. A uniform flow of implementation is devised for both Android and iOS\nsmartphones. The advantage of using multi-threading to achieve or improve real-time throughputs\nis also showcased. A benchmarking framework consisting of accuracy, CPU/GPU consumption,\nand real-time throughput is considered for validation purposes. The developed deployment approach\nallows deep learning models to be turned into real-time smartphone apps with ease based on publicly\navailable deep learning and smartphone software tools. This approach is applied to six popular\nor representative convolutional neural network models, and the validation results based on the\nbenchmarking metrics are reported....
In the signal processing software testing for synthetic aperture radar (SAR), the verification for algorithms is professional and has a\nvery high proportion. However, existing methods can only perform a degree of validation for algorithms, exerting an adverse effect\non the effectiveness of the software testing. This paper proposes a procedure-based approach for algorithm validation. Firstly, it\ndescribes the processing procedures of polar format algorithm (PFA) under the motion-error circumstance, based on which it\nanalyzes the possible questions that may exist in the actual situation. By data simulation, the SAR echoes are generated flexibly and\nefficiently. Then, algorithm simulation is utilized to focus on the demonstrations for the approximations adopted in the algorithm.\nCombined with real data processing, the bugs concealed are excavated further, implementing a comprehensive validation for PFA.\nSimulated experiments and real data processing validate the correctness and effectiveness of the proposed algorithm....
Loading....