With the advent of deep learning, self-driving schemes based on deep learning are becoming more and more popular. Robust perception-action models should learn from data with different scenarios and real behaviors, while current end-to-end model learning is generally limited to training of massive data, innovation of deep network architecture, and learning in-situ model in a simulation environment. Therefore, we introduce a new image style transfer method into data augmentation, and improve the diversity of limited data by changing the texture, contrast ratio and color of the image, and then it is extended to the scenarios that the model has been unobserved before. Inspired by rapid style transfer and artistic style neural algorithms, we propose an arbitrary style generation network architecture, including style transfer network, style learning network, style loss network and multivariate Gaussian distribution function. The style embedding vector is randomly sampled from the multivariate Gaussian distribution and linearly interpolated with the embedded vector predicted by the input image on the style learning network, which provides a set of normalization constants for the style transfer network, and finally realizes the diversity of the image style. In order to verify the effectiveness of the method, image classification and simulation experiments were performed separately. Finally, we built a small-sized smart car experiment platform, and apply the data augmentation technology based on image style transfer drive to the experiment of automatic driving for the first time. The experimental results show that: (1) The proposed scheme can improve the prediction accuracy of the end-to-end model and reduce the model’s error accumulation; (2) the method based on image style transfer provides a new scheme for data augmentation technology, and also provides a solution for the high cost that many deep models rely heavily on a large number of label data.
In this paper, an antiphishing method, called resource request based phishing discovery (RRPD), has been discussed. By analyzing the resources request characteristics of phishing websites, this method can be used on both the client and server sides. On the client side, client RRPD can be used by the web browser for phishing sites detection. On the server side, server RRPD based on the domain name system dataflow can discover the suspicious phishing sites by analyzing a small amount of web content, which saves bandwidth and computing resources to the utmost. Experimental results demonstrate the effectiveness of the proposed methods.
In recent years, phishing has become one of the biggest security threats on the Internet. To combat phishing, it requires multiple steps and multi-agency participation and thus desperately need uniform data sharing format and unobstructed sharing channels, which unfortunately is just what is lacking currently. This paper proposes a novel phishing data sharing mechanism based on the consortium blockchain. It designs four types of nodes, including reporting node, accounting node, servicing node and supervising node and illustrates the roles of each type. Then it demonstrates the process of reporting, accounting and servicing and designs the process of post-supervision, which ensures the operation of the mechanism stable and fastest; and then discusses its implementation on Hyperledger Fabric. The proposed mechanism includes multi-source reporting, anti-tamper accounting, multi-channel disposal of phishing data and post-supervision. It provides a platform for multi-party participation, transparent and efficient coordination and unified standard and overcomes the current prominent problems of phishing data sharing; and the participants on the consortium blockchain all have a strong desire to combat phishing, which ensures the proposed mechanism is also very practical and highly feasible.
To a certain degree, the resilience of the transportation system expresses the safety of the transportation system, because it reflects the ability of the system to maintain its function in the face of disturbance events. In the current research, the assessment of the resilience of urban mobility is attractive and challenging. Apart from this, the concept of green mobility has been popular in recent years. As a representative way of shared mobility, the implementation of ridesharing will affect the level of urban mobility resilience to a certain extent. In this paper, we use a data low-intensity method to evaluate the urban traffic resilience under the circumstance of restricted car use. In addition, we incorporate the impact of ridesharing services. The research in this paper can be regarded as an evaluation framework, which can help policy makers and relevant operators to grasp the overall resilience characteristics of cities in emergencies, identify weak sectors, and formulate the best response plan. This method has been successfully applied to two cities in China, demonstrating its potential for practice. Finally, we also explored the relationship between urban traffic resilience and the pattern of population distribution. The analysis shows that population density has an impact on the level of transportation resilience. And the incorporation of ridesharing will bring an obvious increment in resilience of most areas.
Although a variety of techniques to detect malicious websites have been proposed, it becomes more and more difficult for those methods to provide a satisfying result nowadays. Many malicious websites can still escape detection with various Web spam techniques. In this paper, we first summarize three types of Web spam techniques used by malicious websites, such as redirection spam, hidden IFrame spam, and content hiding spam. We then present a new detection method that adopts the perspective of users and takes screenshots of malicious webpages to invalidate Web spams. The proposed detection method uses a Convolutional Neural Network, which is a class of deep neural networks, as a classification algorithm. In order to verify the effectiveness of the method, two different experiments have been conducted. First, the proposed method was tested based on a constructed complex dataset. We present comparison results between the proposed method and representative machine learning-based detection algorithms. Second, the proposed method was tested to detect malicious websites in a real-world Web environment for three months. These experimental results illustrate that the proposed method has a better performance and is applicable to a practical Web environment.
Abstract Tremendous information is hidden in the light curve of a gamma-ray burst (GRB). Based on Compton Gamma Ray Observatory/Burst And Transient Source Experiment (BATSE) data, Hakkila found a majority of GRBs can be characterized by a smooth, single-peaked component superposed with a temporally symmetrical residual structure, i.e., a mirror feature for the fast-varying component. In this study, we conduct a similar analysis on the same data, as well as on Fermi/Gamma-Ray Burst Monitor data. We obtained a similar conclusion, which is that most GRBs have this symmetrical fast-varying component. Furthermore, we chose an alternative model to characterize the smooth component and used a three-parameter model to identify the residual, i.e., the fast component. By choosing 226 BATSE GRBs based on a few criteria, we checked the time-symmetrical feature and time-translational feature for the fast components and found the ratio is roughly 1:1. We propose that both features could come from the structure of the ejected shells. In the future, the Square Kilometre Array might be able to observe the early radio emission from the collision of the shells.
'/%3e%3cuse xlink:href='%23a' transform='rotate(23.036 .093 25.536)'/%3e%3cuse xlink:href='%23a' transform='rotate(45.87 1.273 16.18)'/%3e%3cuse xlink:href='%23a' transform='rotate(69.945 .996 12.078)'/%3e%3cuse xlink:href='%23a' transform='rotate(20.66 -19.689 31.932)'/%3e%3c/svg%3e)