Olga Ilina

Object detection in images is a fundamental component of many safety-critical systems, such as autonomous driving, video surveillance systems, and robotics. Adversarial patch attacks, being easily implemented in the real world, provide effective counteraction to object detection by state-of-the-art neural-based detectors. It poses a serious danger in various fields of activity. Existing defense methods against patch attacks are insufficiently effective, which underlines the need to develop new reliable solutions. In this manuscript, we propose a method which helps to increase the robustness of neural network systems to the input adversarial images. The proposed method consists of a Deep Convolutional Neural Network to reconstruct a benign image from the adversarial one; a Calculating Maximum Error block to highlight the mismatches between input and reconstructed images; a Localizing Anomalous Fragments block to extract the anomalous regions using the Isolation Forest algorithm from histograms of images’ fragments; and a Clustering and Processing block to group and evaluate the extracted anomalous regions. The proposed method, based on anomaly localization, demonstrates high resistance to adversarial patch attacks while maintaining the high quality of object detection. The experimental results show that the proposed method is effective in defending against adversarial patch attacks. Using the YOLOv3 algorithm with the proposed defensive method for pedestrian detection in the INRIAPerson dataset under the adversarial attacks, the mAP50 metric reaches 80.97% compared to 46.79% without a defensive method. The results of the research demonstrate that the proposed method is promising for improvement of object detection systems security.

The paper presents the results of modeling the impact of changing the ad hoc radio communication network nodes’ adaptive antenna systems radiation pattern (RP) on the quality metrics of the communication protocol network layer. Modeling is carried out using a cross-level approach, which allows to consider various OSI model layers jointly. This paper deals with physical and network layer interaction. Packet transmission, reception and collisions are modeled at the physical layer, contrary to the commonly used approach. Cross-layer modelling provides greater accuracy compared to purely statistical methods of modeling network layer exclusively. We consider an ad hoc radio communication network with identical nodes having antenna systems with variable radiation patterns and making pseudo-random moves in a given region. The simulation method shows that there is an optimal value of the RP width that provides the highest packet receiving probability. Further narrowing of the RP leads to a decrease in the packet receiving probability. The obtained results can be applied to the research, development and optimization of self-organizing wireless communication networks and the equipment for such networks to analyze and control the influence of physical layer parameters on the performance of the communication protocol network layer.

Object detection in images is a fundamental component of many safety-critical systems, such as autonomous driving, video surveillance systems, and robotics. Adversarial patch attacks, which are easily implemented in the real world, provide effective counteraction to the object detection by state-of-the-art neural-based detectors. It poses a serious danger in various fields of activity. Existing defense methods against patch attacks are insufficiently effective, which underlines the need to develop new reliable solutions. In this manuscript, a new approach is proposed to increase the robustness of neural network systems to the input adversarial images. The proposed method, based on the anomaly localization, demonstrates high resistance to adversarial patch attacks while maintaining the high quality of the object detection. The results of the study demonstrate that the proposed method is promising for security of object detection systems improvement and threat of adversarial patch attacks counteraction.

A number of modern techniques for neural network training and recognition enhancement are based on their structures’ symmetry. Such approaches demonstrate impressive results, both for recognition practice, and for understanding of data transformation processes in various feature spaces. This survey examines symmetrical neural network architectures—Siamese and triplet. Among a wide range of tasks having various mathematical formulation areas, especially effective applications of symmetrical neural network architectures are revealed. We systematize and compare different architectures of symmetrical neural networks, identify genetic relationships between significant studies of different authors’ groups, and discuss opportunities to improve the element base of such neural networks. Our survey builds bridges between a large number of isolated studies with significant practical results in the considered area of knowledge, so that the presented survey acquires additional relevance.