In the era of online presence, fake news is widely spread on social media platforms. Fake news refers to false or misleading information that is presented as legitimate news. Unfortunately, a large volume of this fake or misleading information is distributed on the Internet, which is a very serious concern. Among all existing social networks, one of the most rapidly evolving platforms is Threads. This work presents a comprehensive approach to the detection of fake news on Threads using crowd-sourced information. In our detection pipeline, data are initially collected from Threads, with posts labeled by cross-referencing reputable fact-checking platforms such as PolitiFact and Lead Stories. Subsequently, extensive data preprocessing and exploratory analysis are performed to understand engagement patterns, sentiment differences, and posting behaviors. To mitigate the inherent class imbalance between real and fake news posts, synthetic fake news samples are generated using a Large Language Model. For generating robust feature embeddings from the posts, Sentence-BERT is used. We evaluated multiple classification models on the scraped dataset, with Extreme Gradient Boosting (XGBoost) demonstrating the best accuracy of 91.22% (among traditional classifiers) andRoBERTa-baseresulting in 91.93% accuracy (among Transformer-based models). Importantly, we were able to infer multiple interesting data patterns that are unique to Threads (e.g., bot behavior). Hence, this study integrates crowd-sourced data and modern natural language processing (NLP) techniques to develop an accurate fake news detection pipeline. The source code of the project and the FakeThreads dataset are publicly accessible athttps://github.com/Abhi-Ptdr/ThreadsFND.

Data has become an integral part of our digital lives. Specifically, there is an explosive growth in the application of video data. Video information is quite different from other data formats in the sense that it possesses unique characteristics such as high dimensions, complex content, and multiple forms of representation. All these properties make the protection of privacy in videos a complex and challenging task. Simple obscuration techniques cannot address the conclusions or deductions extrapolated from the background information of the entities in the video. In this work, we explore a video sanitization technique that generates synthetic videos following the perturbation of the objects of interest. In our model, we combine the naive detect and obscure technique with randomization in the presence of the objects of interest in each frame and their trajectories. Essentially, our holistic model fulfills the privacy notion of$\epsilon$-object indistinguishability. The generated videos achieve our aim of preserving privacy while being accurate enough for utility analysis. We tested our system on the MOT16 videos dataset and observed a reasonable count of 20% lost objects, mean square error ranging in$[0.2-0.3]$, and trajectories deviation between$[0.2-0.6]$.

A Blockchain network consists of a distributed ledger and a set of nodes participating in the network. The consistency in the ledger’s state is maintained by a Blockchain consensus mechanism. This property is essential since it is assumed that the participants in the network do not trust each other. Most Blockchain consensus algorithms are unsuitable for resource-constrained nodes due to their power consumption, storage, network overhead, and computation requirements. Furthermore, many consensus methods do not address the issue of data accountability, which guarantees that an entity will be judged on its performance or conduct concerning a duty they have taken. This study proposes a novel consensus mechanism termed Proof of Block Height (PoBH) that prioritizes accountability for reporting IoT data transactions and considers factors such as low computation cost and energy consumption. The approach relies on a pre-existing comparable deterministic value based on the height of the committed blocks. The model identifies a malicious intent using a set of predicates and consequently incurs a computation penalty to keep such nodes away from the network. As a preventive measure, the model also seeks to increment this penalty appropriately by making the network free from dishonest nodes for as long as possible, thus sanitizing the network. Importantly, the developed method does not augment the basic hierarchy of an IoT network. A formal analysis of the mechanism’s operation shows that the resource and time required to cheat the system grows exponentially with the number of faulty blocks a node tries to append to the shared ledger. The factors affecting the mechanism’s performance, scalability, and effort required (by malicious nodes) to stay within the network are also quantitatively evaluated.

Several software vulnerabilities emerge during the design phase of a software development process, which can be addressed using secure design patterns. However, using these patterns over web application vulnerabilities is comparatively more tricky for developers than using traditional design patterns. Although several practices exist for addressing software security vulnerabilities, they are sometimes difficult to reuse due to their implementation-specific nature. In this study, we discuss the secure design patterns that are intended to prevent vulnerabilities from being accidentally introduced into code or reduce the effects of flaws. The patterns are created by combining current best security design practices and adding security-specific functionality to the existing design patterns. Hence, this work outlines a convenient mechanism for deciding which secure design patterns to use for addressing online application vulnerabilities. We have demonstrated the applicability of our concept over a prevalent database security threat, namely SQL injection.

Presentation Attacks (PAs) are a common spoofing mechanism in biometric authentications, especially iris-based models. The detection of these attacks is useful for distinguishing whether a sensor is presented with a live biometric or impersonated biometric through a recording, printout or spoof In recent studies, Convolutional Neural Networks have shown exceptional performance in detecting these attacks. In this paper, we propose an attention-based iris PA detection (PAD) termed d-CBAM that uses a convolution block attention mechanism introduced between the dense blocks of DenseNet. The core of this work is inspired by the use of DenseNet as a feature extractor i.e, feature maps from the last dense block are taken and passed to the attention maps. We have tested d-CBAM on the benchmark Clarkson, Notre Dame and NDCLD15 datasets, over which d-CBAM has shown better results in comparison to some traditional PAD solutions such as DenseNet, Spoof Net and Meta-Fusion. The error metrics (APCER and BPCER) were also noted to be competitive with the state-of-the-art.

Internet of Things (IoT) enables physical devices embedded with sensors, software, and other technologies to interoperate and exchange data with other systems over the Internet. Privacy is a huge concern for IoT devices as personal information is constantly being shared through them. Though the best industrial standards like end-to-end encryption are being followed to ensure content-based privacy, contextual privacy concerns still exist. This study focuses on user activity inference attacks, where a passive network observer can infer the private in-home activity of a user by analyzing encrypted IoT traffic metadata. Most of the previous solutions addressing these attacks have either reduced the usability of the devices, increased data overhead, or failed against packet-level signature-based attack scenarios. This study introduces a new defense mechanism that combines dummy packet generation with dynamic link padding. This process makes it difficult for the adversary to avail contextual information about the state of the device (ON or OFF), along with the temporal information (time of state change) from encrypted IoT traffic metadata. We reverse the packet-level signature-based attacks to get device-specific signatures, which helps us generate dummy traffic for the duration of device-specific signatures. Consequently, this results in increased false positives for device state identifications and low traffic overhead. We simulate a state-of-the-art attack scenario to test and vindicate our solution over existing data sets.

Iris-based biometric models are widely recognized to be one of the most accurate forms for authenticating individual identities. Features extracted from the captured iris images (known as IrisCodes) conventionally get stored in their native format over a data repository. However, from a security aspect, the stored templates are highly vulnerable to a wide spectrum of adversarial attack forms. The study in this paper addresses this issue by introducing a privacy-preserving and secure biometric scheme based on the notion of locality sensitive hashing (LSH). In this paper, we have generated cancelable IrisCode features, coined as locality sampled code (LSC), which simultaneously provides strong security guarantees and satisfactory system performance. The functionality of our proposed framework pivots around the fact that intra-class IrisCode samples are “close” to each other, due to which they hash to the same location. Alternatively, the inter-class IrisCodes features are comparatively dissimilar and consequently hash to different locations. We have rigorously examined the intrinsic properties of the LSCs by estimating the intra-class and inter-class collision probabilities for two distinct IrisCodes. Furthermore, we have formally analyzed the security guarantees of non-invertibility, revocability, and unlinkability in our model by establishing various bounds on the adversarial success probability. Extensive empirical tests on the CASIAv3 and IITD benchmark iris databases demonstrate the superior performance of our proposed model, for which we have obtained the best EERs of 0.105% and 1.4%, respectively.