Crowdsensing systems collect various types of data from sensors embedded on mobile devices owned by individuals. These individuals are commonly referred to as workers that complete tasks published crowdsensing systems. Because the relative lack control over worker identities, susceptible poisoning attacks which interfering with analysis results injecting fake conflicting ground truth. Frameworks like TruthFinder can resolve conflicts evaluating trustworthiness providers. frameworks somehow...

5G networks are currently being developed rapidly and dedicated to connect all things of the Internet with new technologies, components services. Due crucial role 5G, as well architectures designs, faces a great number security threats demands suitable technologies. This article proposes an automated attack defense framework toward aim providing exploratory guidance for research. We first review challenges attack/defense objects from perspective hierarchical structure, then propose model...

In softwarized industrial networking, honeypot identification is very important for both the attacker and defender. Existing relies on simple features of honeypot. There exist two challenges: The feature easily simulated, which causes inaccurate results, whereas advanced high interactions, lead to security risks. To cope with these challenges, in this article, we propose a secure fuzzy testing approach inspired by vulnerability mining. It utilizes error handling distinguish honeypots real...

Embodied AI systems, including robots and autonomous vehicles, are increasingly integrated into real-world applications, where they encounter a range of vulnerabilities stemming from both environmental system-level factors. These manifest through sensor spoofing, adversarial attacks, failures in task motion planning, posing significant challenges to robustness safety. Despite the growing body research, existing reviews rarely focus specifically on unique safety security embodied systems....

In this paper, we present Adaptive Computation Steps (ACS) algorithm, which enables end-to-end speech recognition models to dynamically decide how many frames should be processed predict a linguistic output. The model that applies ACS algorithm follows the encoder-decoder framework, while unlike attention-based models, it produces alignments independently at encoder side using correlation between adjacent frames. Thus, predictions can made as soon sufficient acoustic information is received,...

With the widespread deployment of data-driven services, demand for data volumes continues to grow. At present, many applications lack reliable human supervision in process collection, which makes collected contain low-quality or even malicious data. This make AI systems potentially face much security challenges. One main threats training phase machine learning is poisoning attacks, compromise model integrity by contaminating resulting skewed unusable. paper reviews relevant researches on...

Redundant space transfer (RST) encryption can effectively enhance the security and embedding capacity of reversible data hiding in encrypted images. To demonstrate risks RST cloud environment, according to known plaintext image marked image(MEI) pairs, this paper proposes a attack (KPA) based on hamming-weight bit block. The proposed KPA method includes two stages. first stage is find correct ciphertext image. For possible block sizes, all images MEI are reconstructed by deleting embedded...

The recommendation algorithm based on collaborative filtering is vulnerable to data poisoning attacks, wherein attackers can manipulate system output by injecting a large volume of fake rating data. To address this issue, it essential investigate methods for detecting systematically injected within the matrix. Since often inject significant quantity in short period achieve their desired impact, these may exhibit spatial proximity. In other words, be concentrated adjacent rows This paper...

Diabetic retinopathy (DR) is a common complication of diabetes mellitus that poses threat to adults. MicroRNAs (miRNAs) play key role in DR progression. However, the and mechanism miR-192-5p remain unclear. We aimed investigate effect on cell proliferation, migration angiogenesis DR.Expression miR-192-5p, ELAV-like RNA binding protein 1 (ELAVL1) phosphoinositide 3-kinase delta (PI3Kδ) human retinal fibrovascular membrane (FVM) samples microvascular endothelial cells (HRMECs) was assessed...

In the applications of multi-hop wireless sensor networks, effective evaluation link quality which can improve network reliability is a crucial problem in design protocols. The lossy and dynamic nature radio channel, however, makes estimation great challenge. First, we use data-driven modeling statistical inference methods to model burstiness packet loss, reflects retransmission data layer. Then, from multidimensional view characterizing quality, present fuzzy logic based indicator (FLI),...

Transformer-based end-to-end (E2E) automatic speech recognition (ASR) systems have recently gained wide popularity, and are shown to outperform E2E models based on recurrent structures a number of ASR tasks. However, like other models, Transformer also requires the full input sequence for calculating attentions both encoder decoder, leading increased latency posing challenge online ASR. The paper proposes Decoder-end Adaptive Computation Steps (DACS) algorithm address issue facilitate...

There is a large amount of real-time data, e.g., measurement data and instructions, among controllers, sensors, actuators in the Industrial IoT. These are vulnerable to unauthorized access tampering. In addition, once controller controlled by malicious code, it may send out dangerous instruction that not compliant with preset control process, which seriously interferes industrial process. To achieve correct undisturbed based on we propose an attribute-based (ABAC) method for IoT mitigate...

Organ-dedicated PET scanners are becoming more prevalent because of their advantages in higher sensitivity, improved image quality, and lower cost. Detectors utilized these have finer pixel size with depth interaction (DOI) capability. This work presents a LYSO(Ce) detector module DOI capability which has the potential to be scaled up high-resolution small animal or organ-dedicated system. For capability, submodule one LYSO block utilizing PETsys TOFPET2 application-specific integrated...

Knowledge graph (KG) is used to represent the relationships between different concepts in real world. It a special network which nodes entities and edges relationships. KGs can intuitively model connections facts, but many applications, there are certain limitations directly using symbolic logic knowledge perform calculations, making it difficult achieve expected results downstream tasks. Meanwhile, with explosive growth of Internet capacity, traditional KG structure faces problems...

The 5th generation (5G) network adopts a great number of revolutionary technologies to fulfill continuously increasing requirements variety applications, including ultra-high bandwidth, ultra-low latency, ultra-massive device access, ultra-reliability, and so on. Correspondingly, traditional security focuses on the core network, logical (non-physical) layer is no longer suitable for 5G network. presents tendency extend from center edge physical layer. also an essential part security....

In real-world many internet-based service companies need to closely monitor large amounts of data in order ensure stable operation their business. However, anomaly detection for these with various patterns and quality has been a great challenge, especially without labels. this paper, we adopt an algorithm based on Long Short-Term Memory (LSTM) Network terms reconstructing KPIs predicting KPIs. They use the reconstruction error prediction respectively as criteria judging anomalies, test our...