arXiv subjects

Chau-Wai Wong

Publications and source records attributed to Chau-Wai Wong.

At least 19 recordsLinked to original sources

arXiv2026

The Blessing of Dimensionality: How Near-Orthogonality in High-Dimensional Spaces Explains Temporal Portability

Fine-tuning has been widely used to adapt large language models (LLMs) for domain-specific tasks. Parameter efficient fine-tuning (PEFT) methods such as low-rank adaptation (LoRA) are frequently used to reduce computational costs. PortLLM is a training-free and data-free scheme used to adapt LLMs after continual pretraining. Although the initial PortLLM results show that LoRA patches exhibit short-term temporal portability, the long-term performance of PortLLM across several updates of continual pretraining remains underexplored. Furthermore, the intriguing effectiveness of PortLLM is not well understood from a theoretical standpoint. We address these two open questions by (1) performing an extensive empirical study of the long-term temporal portability of PortLLM patches across 10 continual pretraining steps using base models Mistral, Gemma, and Qwen; and (2) offering two theoretical analyses to explain our observation that the simple PortLLM method achieves competitive performance. We find empirically that the portability persists across longer time duration, indicating that repeated fine-tuning is not required when the base model is periodically updated. We find theoretically that near-orthogonality of high-dimensional vectors is a key justification for temporal portability. Our analyses also demonstrate a geometric perspective of the loss landscape in facilitating the theoretical comparison of different adaptation options.

cs.LG↗

arXiv2026

UAV-Based 3D Spectrum Sensing: Insights on Altitude, Bandwidth, Trajectory, and Effective Antenna Patterns on REM Reconstruction

Spectrum sensing and the generation of 3D Radio Environment Maps (REMs) are essential for enabling spectrum sharing within cognitive radio networks. While Uncrewed Aerial Vehicles (UAVs) offer high-mobility 3D sensing, REM accuracy is challenged by dynamic flight behaviors, where fluctuations in UAV speed and direction introduce measurement inconsistencies. Furthermore, the airframe itself impacts the onboard antenna's radiation characteristics. In this paper, using real-world data, we systematically analyze how REM reconstruction accuracy is shaped by three key pillars: physical sensing parameters like altitude and bandwidth, environmental shadowing, and distortions caused by the UAV airframe. First, we benchmark diverse spatial prediction models, including simple Kriging (SK), ordinary Kriging (OK), trans-Gaussian Kriging, and Gaussian process regression (GPR). We demonstrate that while SK and its trans-Gaussian variant are highly accurate at extreme sample sparsity, OK improves as sample size increases, and GPR serves as the most stable overall baseline. Building on this, we propose a novel matrix completion (MC)-assisted GPR framework that enhances REM reconstruction in the presence of non-uniform spatial smoothness. The method operates by decomposing the REM into two distinct layers: a global smooth component and a highly varying local component. Our analysis based on real-world measurements reveals three key findings: 1) REM accuracy and shadowing variance follow a distinct tri-phasic trend as the UAV altitude increases; 2) REM accuracy significantly improves with increased spectrum bandwidth; and 3) antenna pattern calibration from in-field measurements significantly enhances REM accuracy by accounting for the effect of the UAV airframe.

eess.SP↗

arXiv2026

3D Spectrum Awareness for Radio Dynamic Zones Using Kriging and Matrix Completion

Radio Dynamic Zones (RDZs) are geographically defined areas specifically allocated for testing new wireless technologies. It is essential to safeguard the regular spectrum users outside the zones from the interference caused by the deployed equipment within this zone. Previous works have utilized sparse reference signal received power (RSRP) measurements collected by unmanned aerial vehicles (UAVs) to construct a dense 3D radio map through ordinary Kriging. In this work, we illustrate that matrix completion can outperform ordinary Kriging. We partitioned a 2D area of interest into small square grids where each grid corresponds to a single entry of a matrix. The matrix completion algorithm learns the global structure of the radio environment map by leveraging the low-rank property of propagation maps. Additionally, we illustrate that the simple Kriging and trans-Gaussian Kriging yield better results when the density of known measurements is lower. Earlier works of RSRP prediction involved a training dataset at a single altitude. In this work, we also show that performance can be improved by utilizing a combined dataset from multiple altitudes.

eess.SP↗

arXiv2025

Chip-Surface Based Visual Authentication for Integrated Circuits

The rapid development of the semiconductor industry and the ubiquity of electronic devices have led to a significant increase in the counterfeiting of integrated circuits (ICs). This poses a major threat to public health, the banking industry, and military defense sectors that are heavily reliant on electronic systems. The electronic physically unclonable functions (PUFs) are widely used to authenticate IC chips at the unit level. However, electronic PUFs are limited by their requirement for IC chips to be in working status for measurements and their sensitivity to environmental variations. This paper proposes using optical PUFs for IC chip authentication by leveraging the unique microscopic structures of the packaging surface of individual IC chips. The proposed method relies on color images of IC chip surfaces acquired using a flatbed scanner or mobile camera. Our initial study reveals that these consumer-grade imaging devices can capture meaningful physical features from IC chip surfaces. We then propose an efficient, lightweight verification scheme leveraging specular-reflection-based features extracted from videos, achieving an equal error rate (EER) of 0.0008. We conducted factor, sensitivity, and ablation studies to understand the detailed characteristics of the proposed lightweight verification scheme. Our work is the first to apply the optical PUF principle for the authentication of IC chips, synergizing image and video processing with semiconductor chip technology and demonstrating the potential to significantly enhance the security of the semiconductor supply chain.

eess.SP↗

arXiv2025

Exposing Vulnerabilities in Counterfeit Prevention Systems Utilizing Physically Unclonable Surface Features

Counterfeit products pose significant risks to public health and safety through infiltrating untrusted supply chains. Among numerous anti-counterfeiting techniques, leveraging inherent, unclonable microscopic irregularities of paper surfaces is an accurate and cost-effective solution. Prior work of this approach has focused on enabling ubiquitous acquisition of these physically unclonable features (PUFs). However, we will show that existing authentication methods relying on paper surface PUFs may be vulnerable to adversaries, resulting in a gap between technological feasibility and secure real-world deployment. This gap is investigated through formalizing an operational framework for paper-PUF-based authentication. Informed by this framework, we reveal system-level vulnerabilities across both physical and digital domains, designing physical denial-of-service and digital forgery attacks to disrupt proper authentication. The effectiveness of the designed attacks underscores the strong need for security countermeasures for reliable and resilient authentication based on paper PUFs. The proposed framework further facilitates a comprehensive, stage-by-stage security analysis, guiding the design of future counterfeit prevention systems. This analysis delves into potential attack strategies, offering a foundational understanding of how various system components, such as physical features and verification processes, might be exploited by adversaries.

cs.CR↗

arXiv2025

Collection: UAV-Based RSS Measurements from the AFAR Challenge in Digital Twin and Real-World Environments

This paper presents a comprehensive real-world and Digital Twin (DT) dataset collected as part of the AERPAW Find A Rover (AFAR) Challenge, organized by the NSF Aerial Experimentation and Research Platform for Advanced Wireless (AERPAW) testbed and hosted at the Lake Wheeler Field in Raleigh, North Carolina. The AFAR Challenge was a competition involving five finalist university teams, focused on promoting innovation in unmanned aerial vehicle (UAV)-assisted radio frequency (RF) source localization. Participating teams were tasked with designing UAV flight trajectories and localization algorithms to detect the position of a hidden unmanned ground vehicle (UGV), also referred to as a rover, emitting probe signals generated by GNU Radio. The competition was structured to evaluate solutions in a DT environment first, followed by deployment and testing in the AERPAW outdoor wireless testbed. For each team, the UGV was placed at three different positions, resulting in a total of 29 datasets, 15 collected in a DT simulation environment and 14 in a physical outdoor testbed. Each dataset contains time-synchronized measurements of received signal strength (RSS), received signal quality (RSQ), GPS coordinates, UAV velocity, and UAV orientation (roll, pitch, and yaw). Data is organized into structured folders by team, environment (DT and real-world), and UGV location. The dataset supports research in UAV-assisted RF source localization, air-to-ground (A2G) wireless propagation modeling, trajectory optimization, signal prediction, autonomous navigation, and DT validation. With 300k time-synchronized samples from the real-world experiments, the AFAR dataset enables effective training/testing of deep learning (DL) models and supports robust, real-world UAV-based wireless communication and sensing research.

eess.SP↗

arXiv2025

SCALE: Towards Collaborative Content Analysis in Social Science with Large Language Model Agents and Human Intervention

Content analysis breaks down complex and unstructured texts into theory-informed numerical categories. Particularly, in social science, this process usually relies on multiple rounds of manual annotation, domain expert discussion, and rule-based refinement. In this paper, we introduce SCALE, a novel multi-agent framework that effectively $\underline{\textbf{S}}$imulates $\underline{\textbf{C}}$ontent $\underline{\textbf{A}}$nalysis via $\underline{\textbf{L}}$arge language model (LLM) ag$\underline{\textbf{E}}$nts. SCALE imitates key phases of content analysis, including text coding, collaborative discussion, and dynamic codebook evolution, capturing the reflective depth and adaptive discussions of human researchers. Furthermore, by integrating diverse modes of human intervention, SCALE is augmented with expert input to further enhance its performance. Extensive evaluations on real-world datasets demonstrate that SCALE achieves human-approximated performance across various complex content analysis tasks, offering an innovative potential for future social science research.

cs.AI↗

arXiv2025

Advancing Hybrid Defense for Byzantine Attacks in Federated Learning

Federated learning (FL) enables multiple clients to collaboratively train a global model without sharing their local data. Recent studies have highlighted the vulnerability of FL to Byzantine attacks, where malicious clients send poisoned updates to degrade model performance. In particular, many attacks have been developed targeting specific aggregation rules, whereas various defense mechanisms have been designed for dedicated threat models. This paper studies the resilience of attack-agnostic FL scenarios, where the server lacks prior knowledge of both the attackers' strategies and the number of malicious clients involved. We first introduce hybrid defenses against state-of-the-art attacks. Our goal is to identify a general-purpose aggregation rule that performs well on average while also avoiding worst-case vulnerabilities. By adaptively selecting from available defenses, we demonstrate that the server remains robust even when confronted with a substantial proportion of poisoned updates. We also emphasize that existing FL defenses should not automatically be regarded as secure, as demonstrated by the newly proposed Trapsetter attack. The proposed attack outperforms other state-of-the-art attacks by further increasing the impact of the attack by 5-15%. Our findings highlight the ongoing need for the development of Byzantine-resilient aggregation algorithms in FL.

cs.DC↗

arXiv2025

Byzantine Outside, Curious Inside: Reconstructing Data Through Malicious Updates

Federated learning (FL) enables decentralized machine learning without sharing raw data, allowing multiple clients to collaboratively learn a global model. However, studies reveal that privacy leakage is possible under commonly adopted FL protocols. In particular, a server with access to client gradients can synthesize data resembling the clients' training data. In this paper, we introduce a novel threat model in FL, named the maliciously curious client, where a client manipulates its own gradients with the goal of inferring private data from peers. This attacker uniquely exploits the strength of a Byzantine adversary, traditionally aimed at undermining model robustness, and repurposes it to facilitate data reconstruction attack. We begin by formally defining this novel client-side threat model and providing a theoretical analysis that demonstrates its ability to achieve significant reconstruction success during FL training. To demonstrate its practical impact, we further develop a reconstruction algorithm that combines gradient inversion with malicious update strategies. Our analysis and experimental results reveal a critical blind spot in FL defenses: both server-side robust aggregation and client-side privacy mechanisms may fail against our proposed attack. Surprisingly, standard server- and client-side defenses designed to enhance robustness or privacy may unintentionally amplify data leakage. Compared to the baseline approach, a mistakenly used defense may instead improve the reconstructed image quality by 10-15%.

cs.LG↗

arXiv2025

NTK-DFL: Enhancing Decentralized Federated Learning in Heterogeneous Settings via Neural Tangent Kernel

Decentralized federated learning (DFL) is a collaborative machine learning framework for training a model across participants without a central server or raw data exchange. DFL faces challenges due to statistical heterogeneity, as participants often possess data of different distributions reflecting local environments and user behaviors. Recent work has shown that the neural tangent kernel (NTK) approach, when applied to federated learning in a centralized framework, can lead to improved performance. We propose an approach leveraging the NTK to train client models in the decentralized setting, while introducing a synergy between NTK-based evolution and model averaging. This synergy exploits inter-client model deviation and improves both accuracy and convergence in heterogeneous settings. Empirical results demonstrate that our approach consistently achieves higher accuracy than baselines in highly heterogeneous settings, where other approaches often underperform. Additionally, it reaches target performance in 4.6 times fewer communication rounds. We validate our approach across multiple datasets, network topologies, and heterogeneity settings to ensure robustness and generalization.

cs.LG↗

arXiv2025

Federated Learning Nodes Can Reconstruct Peers' Image Data

Federated learning (FL) is a privacy-preserving machine learning framework that enables multiple nodes to train models on their local data and periodically average weight updates to benefit from other nodes' training. Each node's goal is to collaborate with other nodes to improve the model's performance while keeping its training data private. However, this framework does not guarantee data privacy. Prior work has shown that the gradient-sharing steps in FL can be vulnerable to data reconstruction attacks from an honest-but-curious central server. In this work, we show that an honest-but-curious node/client can also launch attacks to reconstruct peers' image data through gradient inversion, presenting a severe privacy risk. We demonstrate that a single client can silently reconstruct other clients' private images using diluted information available within consecutive updates. We leverage state-of-the-art diffusion models to enhance the perceptual quality and recognizability of the reconstructed images, further demonstrating the risk of information leakage at a semantic level. This highlights the need for more robust privacy-preserving mechanisms that protect against silent client-side attacks during federated training.

cs.LG↗

arXiv2025

Individualized Deepfake Detection Exploiting Traces Due to Double Neural-Network Operations

In today's digital landscape, journalists urgently require tools to verify the authenticity of facial images and videos depicting specific public figures before incorporating them into news stories. Existing deepfake detectors are not optimized for this detection task when an image is associated with a specific and identifiable individual. This study focuses on the deepfake detection of facial images of individual public figures. We propose to condition the proposed detector on the identity of an identified individual, given the advantages revealed by our theory-driven simulations. While most detectors in the literature rely on perceptible or imperceptible artifacts present in deepfake facial images, we demonstrate that the detection performance can be improved by exploiting the idempotency property of neural networks. In our approach, the training process involves double neural-network operations where we pass an authentic image through a deepfake simulating network twice. Experimental results show that the proposed method improves the area under the curve (AUC) from 0.92 to 0.94 and reduces its standard deviation by 17%. To address the need for evaluating detection performance for individual public figures, we curated and publicly released a dataset of ~32k images featuring 45 public figures, as existing deepfake datasets do not meet this criterion.

eess.IV↗

arXiv2025

HATFormer: Historic Handwritten Arabic Text Recognition with Transformers

Arabic handwritten text recognition (HTR) is challenging, especially for historical texts, due to diverse writing styles and the intrinsic features of Arabic script. Additionally, Arabic handwriting datasets are smaller compared to English ones, making it difficult to train generalizable Arabic HTR models. To address these challenges, we propose HATFormer, a transformer-based encoder-decoder architecture that builds on a state-of-the-art English HTR model. By leveraging the transformer's attention mechanism, HATFormer captures spatial contextual information to address the intrinsic challenges of Arabic script through differentiating cursive characters, decomposing visual representations, and identifying diacritics. Our customization to historical handwritten Arabic includes an image processor for effective ViT information preprocessing, a text tokenizer for compact Arabic text representation, and a training pipeline that accounts for a limited amount of historic Arabic handwriting data. HATFormer achieves a character error rate (CER) of 8.6% on the largest public historical handwritten Arabic dataset, with a 51% improvement over the best baseline in the literature. HATFormer also attains a comparable CER of 4.2% on the largest private non-historical dataset. Our work demonstrates the feasibility of adapting an English HTR method to a low-resource language with complex, language-specific challenges, contributing to advancements in document digitization, information retrieval, and cultural preservation.

cs.CV↗

arXiv2025

UAV-Assisted Coverage Hole Detection Using Reinforcement Learning in Urban Cellular Networks

Deployment of cellular networks in urban areas requires addressing various challenges. For example, high-rise buildings with varying geometrical shapes and heights contribute to signal attenuation, reflection, diffraction, and scattering effects. This creates a high possibility of coverage holes (CHs) within the proximity of the buildings. Detecting these CHs is critical for network operators to ensure quality of service, as customers in these areas may experience weak or no signal reception. To address this challenge, we propose an approach using an autonomous vehicle, such as an unmanned aerial vehicle (UAV), to detect CHs, for minimizing drive test efforts and reducing human labor. The UAV leverages reinforcement learning (RL) to find CHs using stored local building maps, its current location, and measured signal strengths. As the UAV moves, it dynamically updates its knowledge of the signal environment and its direction to a nearby CH while avoiding collisions with buildings. We created a wide range of testing scenarios using building maps from OpenStreetMap and signal strength data generated by NVIDIA Sionna raytracing simulations. The results show that the RL-based approach outperforms non-machine learning, geometry-based methods in detecting CHs in urban areas. Additionally, even with a limited number of UAV measurements, the method achieves performance close to theoretical upper bounds that assume complete knowledge of all signal strengths.

eess.SP↗

arXiv2025

PortLLM: Personalizing Evolving Large Language Models with Training-Free and Portable Model Patches

As large language models (LLMs) increasingly shape the AI landscape, fine-tuning pretrained models has become more popular than in the pre-LLM era for achieving optimal performance in domain-specific tasks. However, pretrained LLMs such as ChatGPT are periodically evolved, i.e., model parameters are frequently updated), making it challenging for downstream users with limited resources to keep up with fine-tuning the newest LLMs for their domain application. Even though fine-tuning costs have nowadays been reduced thanks to the innovations of parameter-efficient fine-tuning such as LoRA, not all downstream users have adequate computing for frequent personalization. Moreover, access to fine-tuning datasets, particularly in sensitive domains such as healthcare, could be time-restrictive, making it crucial to retain the knowledge encoded in earlier fine-tuned rounds for future adaptation. In this paper, we present PortLLM, a training-free framework that (i) creates an initial lightweight model update patch to capture domain-specific knowledge, and (ii) allows a subsequent seamless plugging for the continual personalization of evolved LLM at minimal cost. Our extensive experiments cover seven representative datasets, from easier question-answering tasks {BoolQ, SST2} to harder reasoning tasks {WinoGrande, GSM8K}, and models including {Mistral-7B, Llama2, Llama3.1, and Gemma2}, validating the portability of our designed model patches and showcasing the effectiveness of our proposed framework. For instance, PortLLM achieves comparable performance to LoRA fine-tuning with reductions of up to 12.2x in GPU memory usage. Finally, we provide theoretical justifications to understand the portability of our model update patches, which offers new insights into the theoretical dimension of LLMs' personalization.

cs.CL↗

arXiv2025

Muharaf: Manuscripts of Handwritten Arabic Dataset for Cursive Text Recognition

We present the Manuscripts of Handwritten Arabic~(Muharaf) dataset, which is a machine learning dataset consisting of more than 1,600 historic handwritten page images transcribed by experts in archival Arabic. Each document image is accompanied by spatial polygonal coordinates of its text lines as well as basic page elements. This dataset was compiled to advance the state of the art in handwritten text recognition (HTR), not only for Arabic manuscripts but also for cursive text in general. The Muharaf dataset includes diverse handwriting styles and a wide range of document types, including personal letters, diaries, notes, poems, church records, and legal correspondences. In this paper, we describe the data acquisition pipeline, notable dataset features, and statistics. We also provide a preliminary baseline result achieved by training convolutional neural networks using this data.

cs.CV↗

arXiv2025

Remote Blood Oxygen Estimation From Videos Using Neural Networks

Blood oxygen saturation (SpO$_2$) is an essential indicator of respiratory functionality and is receiving increasing attention during the COVID-19 pandemic. Clinical findings show that it is possible for COVID-19 patients to have significantly low SpO$_2$ before any obvious symptoms. The prevalence of cameras has motivated researchers to investigate methods for monitoring SpO$_2$ using videos. Most prior schemes involving smartphones are contact-based: They require a fingertip to cover the phone's camera and the nearby light source to capture re-emitted light from the illuminated tissue. In this paper, we propose the first convolutional neural network based noncontact SpO$_2$ estimation scheme using smartphone cameras. The scheme analyzes the videos of a participant's hand for physiological sensing, which is convenient and comfortable, and can protect their privacy and allow for keeping face masks on. We design our neural network architectures inspired by the optophysiological models for SpO$_2$ measurement and demonstrate the explainability by visualizing the weights for channel combination. Our proposed models outperform the state-of-the-art model that is designed for contact-based SpO$_2$ measurement, showing the potential of our proposed method to contribute to public health. We also analyze the impact of skin type and the side of a hand on SpO$_2$ estimation performance.

cs.LG↗

arXiv2025

A Multi-Channel Ratio-of-Ratios Method for Noncontact Hand Video Based SpO$_2$ Monitoring Using Smartphone Cameras

Blood oxygen saturation (SpO$_2$) is an important indicator for pulmonary and respiratory functionalities. Clinical findings on COVID-19 show that many patients had dangerously low blood oxygen levels not long before conditions worsened. It is therefore recommended, especially for the vulnerable population, to regularly monitor the blood oxygen level for precaution. Recent works have investigated how ubiquitous smartphone cameras can be used to infer SpO$_2$. Most of these works are contact-based, requiring users to cover a phone's camera and its nearby light source with a finger to capture reemitted light from the illuminated tissue. Contact-based methods may lead to skin irritation and sanitary concerns, especially during a pandemic. In this paper, we propose a noncontact method for SpO$_2$ monitoring using hand videos acquired by smartphones. Considering the optical broadband nature of the red (R), green (G), and blue (B) color channels of the smartphone cameras, we exploit all three channels of RGB sensing to distill the SpO$_2$ information beyond the traditional ratio-of-ratios (RoR) method that uses only two wavelengths. To further facilitate an accurate SpO$_2$ prediction, we design adaptive narrow bandpass filters based on accurately estimated heart rate to obtain the most cardiac-related AC component for each color channel. Experimental results show that our proposed blood oxygen estimation method can reach a mean absolute error of 1.26% when a pulse oximeter is used as a reference, outperforming the traditional RoR method by 25%.

eess.SP↗