Wireless sensor networks (WSNs) deployed in hostile environments are vulnerable to clone attacks. In such attack, an adversary compromises a few nodes, replicates them, and inserts arbitrary number of replicas into the network. Consequently, can carry out many internal Previous solutions on detecting attacks have several drawbacks. First, some them require central control, which introduces inherent limits. Second, deterministic simple witness compromising Third, easily learn critical nodes...

Deep learning techniques have made tremendous progress in a variety of challenging tasks, such as image recognition and machine translation, during the past decade. Training deep neural networks is computationally expensive requires both human intellectual resources. Therefore, it necessary to protect property model externally verify ownership model. However, previous studies either fail defend against evasion attack or not explicitly dealt with fraudulent claims by adversaries. Furthermore,...

Deep learning techniques have made tremendous progress in a variety of challenging tasks, such as image recognition and machine translation, during the past decade. Training deep neural networks is computationally expensive requires both human intellectual resources. Therefore, it necessary to protect property model externally verify ownership model. However, previous studies either fail defend against evasion attack or not explicitly dealt with fraudulent claims by adversaries. Furthermore,...

In real life, one requires signatures from people who satisfy certain criteria like that they should possess some specific attributes. For example, Alice wants a document to be signed by employee in Bob's company. This must have attributes such as being part of the IT staff and at least junior manager cryptography team or senior biometrics team. order these kinds needs, we defined common attribute-based signature scheme where signing member has belong group, also proved our secure.

Fully homomorphic encryption (FHE) is a promising data privacy solution for machine learning, which allows the inference to be performed with encrypted data. However, it typically leads 5-6 orders of magnitude higher computation and storage overhead. This paper proposes first full-fledged FPGA acceleration framework FHE-based convolution neural network (HE-CNN) inference. We then design parameterized HE operation modules intra- inter- HE-CNN layer resource management based on high-level...

The rapid advancement of generative models has led to the proliferation highly realistic AI-generated images, posing significant challenges for detection methods generalize across diverse and evolving techniques. Existing approaches often fail adapt unknown without costly retraining, limiting their practicability. To fill this gap, we propose Post-hoc Distribution Alignment (PDA), a novel approach generalizable images. key idea is use known model regenerate undifferentiated test This process...

DeepFakes pose a significant threat to our society. One representative DeepFake application is face-swapping, which replaces the identity in facial image with that of victim. Although existing methods partially mitigate these risks by degrading quality swapped images, they often fail disrupt transformation effectively. To fill this gap, we propose FaceSwapGuard (FSG), novel black-box defense mechanism against deepfake face-swapping threats. Specifically, FSG introduces imperceptible...

Fine-grained hashing is a new topic in the field of hashing-based retrieval and has not been well explored up to now. In this paper, we raise three key issues that fine-grained should address simultaneously, i.e., feature extraction, refinement as well-designed loss function. order these issues, propose novel Fine-graIned haSHing method with double-filtering mechanism proxy-based function, FISH for short. Specifically, consists two modules, Space Filtering module Feature module, which...

Mobile apps have become popular for providing artificial intelligence (AI) services via on-device machine learning (ML) techniques. Unlike accomplishing these AI on remote servers traditionally, techniques process sensitive information required by locally, which can mitigate the severe concerns of data collection side. However, to push core ML expertise (e.g., models) smartphones are still subject similar vulnerabilities clouds and servers, especially when facing model stealing attack. To...

Data privacy becomes a crucial concern in the AI and big data era. Fully homomorphic encryption (FHE) is promising protection technique where entire computation performed on encrypted data. However, dramatic increase of workload restrains usage FHE for real-world applications. In this paper, we propose an FPFA accelerator design framework CKKS-based HE. While KeySwitch operations are primary performance bottleneck computation, low latency module with reduced intra-operation dependency....

With the wide use of Automatic Speech Recognition (ASR) in applications such as human machine interaction, simultaneous interpretation, audio transcription, and so on, its security protection becomes increasingly important. Although recent studies have brought to light weaknesses popular ASR systems that enable out-of-band signal attack, adversarial further proposed various remedies (signal smoothing, training, etc.), a systematic understanding (both attacks defenses) is still missing,...

Recently, multimodal hashing techniques have received considerable attention due to their low storage cost and fast query speed for data retrieval. Many methods been proposed; however, there are still some problems that need be further considered. For example, of these just use a similarity matrix learning hash functions which will discard useful information contained in original data; them relax binary constraints or separate the process codes into two independent stages bypass obstacle...

The publish/subscribe model has gained prominence in the Internet of things (IoT) network, and both Message Queue Telemetry Transport (MQTT) Constrained Application Protocol (CoAP) support it. However, existing coverage-based fuzzers may miss some paths when fuzzing such protocols, because they implicitly assume that there are only two parties a protocol, which is not true now since three parties, i.e., publisher, subscriber broker. In this paper, we propose MultiFuzz, new...

Permission is the fundamental security mechanism for protecting user data and privacy on Android. Given its importance, researchers have studied design usage of permissions from various aspects. However, most previous research focused issues system permissions. Overlooked by many researchers, an app can use custom to share resources capabilities with other apps. implications using not been fully understood.In this paper, we systematically evaluate implementation Android Notably, built...

Global darknet monitoring provides an effective way to observe cyber-attacks that are significantly threatening network security and management. In this paper, we present a study on characterization of cyberattacks in the big stream data collected large scale distributed using association rule learning. The experiment shows learning can support strategic cyberattack countermeasure following ways. First, statistics computed from malware-specific rules lead better understanding global trend...

In light of the rapid growth malware threats towards Android platform, there is a pressing need to develop effective solutions. this paper we explorate potential multi-modal features enhance detection accuracy while keep false alarms low. Examined include permissions, Application Programming Interface (API) calls, and meta such as category information Package (APK) descriptions. These are coded in way facilitate efficient learning testing with particular classifiers known linear support...

The Secure Sockets Layer (SSL) and Transport Security (TLS) protocols are the foundation of network security. certificate verification in SSL/TLS implementations is vital may become "weak link" whole ecosystem. In previous works, some research focused on automated testing verification, main approaches rely generating massive certificates through randomly combining parts seed for fuzzing. Although generated could meet semantic constraints, cost quite heavy, performance limited due to...

The rapid development of Android apps primarily benefits from third-party libraries that provide well-encapsulated functionalities. On the other hand, more and malicious are discovered in wild, which brings new security challenges. Despite some previous studies focusing on libraries, however, most them only study specific types or individual cases. community still lacks a comprehensive understanding potentially (PMLs) wild.