Anupam Das

Ummay Kulsum, Aafaq Sabir, Abhinaya S. B. et al.

YouTube is a major platform for information and entertainment, but its wide accessibility also makes it attractive for scammers to upload deceptive or malicious content. Prior detection approaches rely largely on textual or statistical metadata, such as titles, descriptions, view counts, or likes, which are effective in many cases but can be evaded through benign-looking text, manipulated statistics, or other obfuscation strategies (e.g., 'Leetspeak'), while ignoring visual cues. In this study, we systematically investigate multimodal approaches for detecting YouTube scams. Our dataset consolidates established scam categories and augments them with full-length videos and policy-grounded reasoning annotations. Experiments show that a text-only model using titles and descriptions (fine-tuned BERT) achieves moderate performance (76.61% F1 score), improving slightly with audio transcripts (77.98% F1 score). Visual analysis with a fine-tuned LLaVA-Video model performs better (79.61% F1 score), while a multimodal framework combining titles, descriptions, and video frames achieves the highest performance (82.96% F1 score). Moreover, the multimodal framework showed greater robustness to adversarial perturbations, with accuracy dropping only 1-3%, compared to 12-38% for modality-specific models. Beyond accuracy, the multimodal framework provides interpretable, policy-grounded reasoning, enhancing transparency and practical utility in automated moderation. Using this approach, we analyzed 6,374 in-the-wild YouTube videos and detected 1,864 scams with explicit reasoning, providing a valuable resource for future research.

Md Adnan Arefeen, Zhouyu Li, Md Yusuf Sarwar Uddin et al.

With the growth of computer vision applications, deep learning, and edge computing contribute to ensuring practical collaborative intelligence (CI) by distributing the workload among edge devices and the cloud. However, running separate single-task models on edge devices is inefficient regarding the required computational resource and time. In this context, multi-task learning allows leveraging a single deep learning model for performing multiple tasks, such as semantic segmentation and depth estimation on incoming video frames. This single processing pipeline generates common deep features that are shared among multi-task modules. However, in a collaborative intelligence scenario, generating common deep features has two major issues. First, the deep features may inadvertently contain input information exposed to the downstream modules (violating input privacy). Second, the generated universal features expose a piece of collective information than what is intended for a certain task, in which features for one task can be utilized to perform another task (violating task privacy). This paper proposes a novel deep learning-based privacy-cognizant feature generation process called MetaMorphosis that limits inference capability to specific tasks at hand. To achieve this, we propose a channel squeeze-excitation based feature metamorphosis module, Cross-SEC, to achieve distinct attention of all tasks and a de-correlation loss function with differential-privacy to train a deep learning model that produces distinct privacy-aware features as an output for the respective tasks. With extensive experimentation on four datasets consisting of diverse images related to scene understanding and facial attributes, we show that MetaMorphosis outperforms recent adversarial learning and universal feature generation methods by guaranteeing privacy requirements in an efficient way for image and video analytics.

Anupam Das, Nikita Borisov, Matthew Caesar

Modern smartphones contain motion sensors, such as accelerometers and gyroscopes. These sensors have many useful applications; however, they can also be used to uniquely identify a phone by measuring anomalies in the signals, which are a result from manufacturing imperfections. Such measurements can be conducted surreptitiously in the browser and can be used to track users across applications, websites, and visits. We analyze techniques to mitigate such device fingerprinting either by calibrating the sensors to eliminate the signal anomalies, or by adding noise that obfuscates the anomalies. To do this, we first develop a highly accurate fingerprinting mechanism that combines multiple motion sensors and makes use of (inaudible) audio stimulation to improve detection. We then collect measurements from a large collection of smartphones and evaluate the impact of calibration and obfuscation techniques on the classifier accuracy.

Anupam Das, Nikita Borisov, Matthew Caesar

The widespread use of smart devices gives rise to both security and privacy concerns. Fingerprinting smart devices can assist in authenticating physical devices, but it can also jeopardize privacy by allowing remote identification without user awareness. We propose a novel fingerprinting approach that uses the microphones and speakers of smart phones to uniquely identify an individual device. During fabrication, subtle imperfections arise in device microphones and speakers which induce anomalies in produced and received sounds. We exploit this observation to fingerprint smart devices through playback and recording of audio samples. We use audio-metric tools to analyze and explore different acoustic features and analyze their ability to successfully fingerprint smart devices. Our experiments show that it is even possible to fingerprint devices that have the same vendor and model; we were able to accurately distinguish over 93% of all recorded audio clips from 15 different units of the same model. Our study identifies the prominent acoustic features capable of fingerprinting devices with high success rate and examines the effect of background noise and other variables on fingerprinting accuracy.