Deep Residual Networks have recently been shown to significantly improve the performance of neural networks trained on ImageNet, with results beating all previous methods this dataset by large margins in image classification task. However, meaning these impressive numbers and their implications for future research are not fully understood yet. In survey, we will try explain what are, how they achieve excellent results, why successful implementation practice represents a significant advance...

Different from ship detection synthetic aperture radar (SDSAR) and spaceborne optical images (SDSOI), visual image (SDVI) has better accuracy real-time performance, which can be widely used in port management, cross-border detection, autonomous ship, safe navigation, other applications. In this paper, we proposed a new SDVI algorithm, named enhanced YOLO v3 tiny network for detection. The algorithm video surveillance to realize the accurate classification positioning of six types ships...

This paper provides an extensive and complete survey on the process of detecting preventing various types IoT-based security attacks. It is designed for software developers, researchers, practitioners in Internet Things field who aim to understand these For each entry identified from list, a brief description provided along with references where more information can be found. However, We surveyed current state-of-the-art IoT solutions focused four main aspects: (1) handpicking representative...

Group recommendation has attracted significant research efforts for its importance in benefiting a group of users. This paper investigates the Recommendation problem from novel aspect, which tries to maximize satisfaction each member while minimizing unfairness between them. In this work, we present several semantics individual utility and propose two concepts social welfare fairness modeling overall utilities balance members. We formulate as multiple objective optimization show that it is...

With the fast development of Internet Things (IoT) technologies, more IoT devices are currently connected with Internet, resulting in exchange information. However, data privacy and security threats have become emerging challenges IoT. In this paper, we concerned about image transmission green Image encryption algorithms for faced two challenges: 1) To save cost, always very low computing power, so they cannot support high-precision 2) The algorithm deployed on device should be efficient to...

Convolutional neural networks, in which each layer receives features from the previous layer(s) and then aggregates/abstracts higher level them, are widely adopted for image classification. To avoid information loss during feature aggregation/abstraction fully utilize lower features, we propose a novel decision fusion module (DFM) making an intermediate based on current fuse its results with original before passing them to next layers. This is devised determine auxiliary category...

Rendezvous is a fundamental process in Cognitive Radio Networks, through which user establishes link to communicate with neighbor on common channel. Most previous solutions use either central controller or Common Control Channel (CCC) simplify the problem, are inflexible and vulnerable faults attacks. Some blind rendezvous algorithms have been proposed that rely no centralization. Hopping (CH) representative technique used rendezvous, each hops among available channels according pre-defined...

Deep neural networks (DNNs) have been widely adopted but they are vulnerable to intentionally crafted adversarial examples. Various attack methods against DNNs proposed, yet there still lacks theoretical explanation of In this paper, we aim understand examples from the attacking process and assume adding perturbations key/sensitive regions image could fool classification DNNs. We propose gradient shielding verify assumption which ignores insensitive information during generating...

The COVID-19 pandemic has caused serious consequences in the last few months and trying to control it been most important objective. With effective prevention methods, epidemic gradually under some countries is essential ensure safe work resumption future. Although approaches are proposed measure people's healthy conditions, such as filling health information forms or evaluating travel records, they cannot provide a fine-grained assessment of risk. In this paper, we propose novel risk method...

Speaker recognition is a task that identifies the speaker from multiple audios. Recently, advances in deep learning have considerably boosted development of speech signal processing techniques. or has been widely adopted such applications as smart locks, vehicle-mounted systems, and financial services. However, neural network-based systems (SRSs) are susceptible to adversarial attacks, which fool system make wrong decisions by small perturbations, this drawn attention researchers security...

Due to globalization and advances in network technology, the Internet of Vehicles (IoV) with edge computing has gained increasingly more attention over last few years. The technology provides a new paradigm design interconnected distributed nodes Unmanned Aerial Vehicle (UAV)-assisted vehicle networks for communications between vehicles smart cities. process hierarchy current UAV-assisted is also becoming multifaceted as are connected, requiring accessing exchanging information, performing...

Adversarial attacks have been successfully extended to the field of point clouds. Besides applying common perturbation guided by gradient, adversarial on clouds can be conducted directional perturbations, e.g., along normal and tangent plane. In this article, we first investigate whether with these two orthogonal perturbations are more imperceptible than that gradient-aware perturbation. Second, deeper difference between they applicable same scenarios. Third, based verification results above...

With the rapid development of information technologies, security cyberspace has become increasingly serious. Network intrusion detection is a practical scheme to protect network systems from cyber attacks. However, as new vulnerabilities and unknown attack types are constantly emerging, only few samples such attacks can be captured for analysis, which cannot handled by existing methods deployed in real systems. To handle this problem, we propose few-shot class-incremental learning method...

Deep neural networks (DNNs) can improve the image analysis and interpretation of remote sensing technology by extracting valuable information from images, has extensive applications such as military affairs, agriculture, environment, transportation, urban division. The DNNs for object detection identify analyze objects in images through fruitful features which improves efficiency processing enables recognition large-scale images. However, many studies have shown that deep are vulnerable to...

The Controller Area Network (CAN) is a bus protocol widely used in intelligent connected vehicles for communication between electronic and systems. However, the continuous increase inter- intra-vehicle traffic makes CAN vulnerable to cyber-attacks, including unknown attacks that have never been seen before. Previous studies either use closed set scenarios misclassify as known classes with high confidence, or models calculate thresholds identify ignoring relationship feature representation...

Graph Retrieval Augmented Generation (GRAG) is a novel paradigm that takes the naive RAG system step further by integrating graph information, such as knowledge (KGs), into large-scale language models (LLMs) to mitigate hallucination. However, existing GRAG still encounter limitations: 1) simple paradigms usually fail with complex problems due narrow and shallow correlations capture from KGs 2) methods of strong coupling tend be high computation cost time consuming if dense. In this paper,...

Graph Retrieval Augmented Generation (GRAG) is a novel paradigm that takes the naive RAG system step further by integrating graph information, such as knowledge (KGs), into large-scale language models (LLMs) to mitigate hallucination. However, existing GRAG still encounter limitations: 1) simple paradigms usually fail with complex problems due narrow and shallow correlations capture from KGs 2) methods of strong coupling tend be high computation cost time consuming if dense. In this paper,...