arXiv subjects

Yasir Zaki

Publications and source records attributed to Yasir Zaki.

At least 37 records · Page 2Linked to original sources

arXiv2025

Towards Next Generation Immersive Applications in 5G Environments

The Multi-user Immersive Reality (MIR) landscape is evolving rapidly, with applications spanning virtual collaboration, entertainment, and training. However, wireless network limitations create a critical bottleneck, struggling to meet the high-bandwidth and ultra-low latency demands essential for next-generation MIR experiences. This paper presents Hera, a modular framework for next-generation immersive applications, comprising a high-level streaming and synchronization layer for AR/VR systems and a low-level delay-based QoE-aware rate control protocol optimized for dynamic wireless environments. The Hera framework integrates application-aware streaming logic with a QoE-centric rate control core, enabling adaptive video quality, multi-user fairness, and low-latency communication across challenging 5G network conditions. We demonstrate that Hera outperforms existing state-of-the-art rate control algorithms by maintaining up to 66% lower latencies with comparable throughput performance, higher visual quality with 50% average bitrate improvements in our analysis, and improved fairness. By bridging the gap between application-level responsiveness and network-level adaptability, Hera lays the foundation for more scalable, robust, and high-fidelity multi-user immersive experiences.

cs.NI↗

arXiv2025

Self-Regulating Cars: Automating Traffic Control in Free Flow Road Networks

Free-flow road networks, such as suburban highways, are increasingly experiencing traffic congestion due to growing commuter inflow and limited infrastructure. Traditional control mechanisms, such as traffic signals or local heuristics, are ineffective or infeasible in these high-speed, signal-free environments. We introduce self-regulating cars, a reinforcement learning-based traffic control protocol that dynamically modulates vehicle speeds to optimize throughput and prevent congestion, without requiring new physical infrastructure. Our approach integrates classical traffic flow theory, gap acceptance models, and microscopic simulation into a physics-informed RL framework. By abstracting roads into super-segments, the agent captures emergent flow dynamics and learns robust speed modulation policies from instantaneous traffic observations. Evaluated in the high-fidelity PTV Vissim simulator on a real-world highway network, our method improves total throughput by 5%, reduces average delay by 13%, and decreases total stops by 3% compared to the no-control setting. It also achieves smoother, congestion-resistant flow while generalizing across varied traffic patterns, demonstrating its potential for scalable, ML-driven traffic management.

cs.LG↗

arXiv2025

SONIC: Cost-Effective Web Access for Developing Countries

Over 2.6 billion people remain without access to the Internet in 2025. This phenomenon is especially pronounced in developing regions, where cost and infrastructure limitations are major barriers to connectivity. In response, we design SONIC, a low-cost, scalable data delivery system that builds on existing infrastructures: FM radio for downlink broadcasting, and SMS for personalized uplink. SONIC is motivated by the widespread availability of FM radio and SMS infrastructure in developing regions, along with embedded FM radio tuners in affordable mobile phones. SONIC offers several innovations to effectively transmit Web content over sound over FM radio, in a reliable and compressed form. For example, we transmit pre-rendered webpages and leverage pixel interpolation to recover errors at the receiver. We further modify Android to offer a simpler deployment pipeline, supporting a wide range of devices. We deployed SONIC at an FM radio station in Cameroon for six weeks with 30 participants. Our results demonstrate a sustained downlink throughput of 10 kbps, less than 20% loss for a majority of transmissions with signal strength above -90 dbM, and a strong user engagement across both Web browsing and ChatGPT interactions.

cs.NI↗

arXiv2025

Data Doping or True Intelligence? Evaluating the Transferability of Injected Knowledge in LLMs

As the knowledge of large language models (LLMs) becomes outdated over time, there is a growing need for efficient methods to update them, especially when injecting proprietary information. Our study reveals that comprehension-intensive fine-tuning tasks (e.g., question answering and blanks) achieve substantially higher knowledge retention rates (48%) compared to mapping-oriented tasks like translation (17%) or text-to-JSON conversion (20%), despite exposure to identical factual content. We demonstrate that this pattern persists across model architectures and follows scaling laws, with larger models showing improved retention across all task types. However, all models exhibit significant performance drops when applying injected knowledge in broader contexts, suggesting limited semantic integration. These findings show the importance of task selection in updating LLM knowledge, showing that effective knowledge injection relies not just on data exposure but on the depth of cognitive engagement during fine-tuning.

cs.CL↗

arXiv2025

TikTok's recommendations skewed towards Republican content during the 2024 U.S. presidential race

TikTok is a major force among social media platforms with over a billion monthly active users worldwide and 170 million in the United States. The platform's status as a key news source, particularly among younger demographics, raises concerns about its potential influence on politics in the U.S. and globally. Despite these concerns, there is scant research investigating TikTok's recommendation algorithm for political biases. We fill this gap by conducting 323 independent algorithmic audit experiments testing partisan content recommendations in the lead-up to the 2024 U.S. presidential elections. Specifically, we create hundreds of "sock puppet" TikTok accounts in Texas, New York, and Georgia, seeding them with varying partisan content and collecting algorithmic content recommendations for each of them. Collectively, these accounts viewed ~394,000 videos from April 30th to November 11th, 2024, which we label for political and partisan content. Our analysis reveals significant asymmetries in content distribution: Republican-seeded accounts received ~11.8% more party-aligned recommendations compared to their Democratic-seeded counterparts, and Democratic-seeded accounts were exposed to ~7.5% more opposite-party recommendations on average. These asymmetries exist across all three states and persist when accounting for video- and channel-level engagement metrics such as likes, views, shares, comments, and followers, and are driven primarily by negative partisanship content. Our findings provide insights into the inner workings of TikTok's recommendation algorithm during a critical election period, raising fundamental questions about platform neutrality.

cs.SI↗

arXiv2025

Neutralizing the Narrative: AI-Powered Debiasing of Online News Articles

Bias in news reporting significantly impacts public perception, particularly regarding crime, politics, and societal issues. Traditional bias detection methods, predominantly reliant on human moderation, suffer from subjective interpretations and scalability constraints. Here, we introduce an AI-driven framework leveraging advanced large language models (LLMs), specifically GPT-4o, GPT-4o Mini, Gemini Pro, Gemini Flash, Llama 8B, and Llama 3B, to systematically identify and mitigate biases in news articles. To this end, we collect an extensive dataset consisting of over 30,000 crime-related articles from five politically diverse news sources spanning a decade (2013-2023). Our approach employs a two-stage methodology: (1) bias detection, where each LLM scores and justifies biased content at the paragraph level, validated through human evaluation for ground truth establishment, and (2) iterative debiasing using GPT-4o Mini, verified by both automated reassessment and human reviewers. Empirical results indicate GPT-4o Mini's superior accuracy in bias detection and effectiveness in debiasing. Furthermore, our analysis reveals temporal and geographical variations in media bias correlating with socio-political dynamics and real-world events. This study contributes to scalable computational methodologies for bias mitigation, promoting fairness and accountability in news reporting.

cs.CL↗

arXiv2025

HowkGPT: Investigating the Detection of ChatGPT-generated University Student Homework through Context-Aware Perplexity Analysis

As the use of Large Language Models (LLMs) in text generation tasks proliferates, concerns arise over their potential to compromise academic integrity. The education sector currently tussles with distinguishing student-authored homework assignments from AI-generated ones. This paper addresses the challenge by introducing HowkGPT, designed to identify homework assignments generated by AI. HowkGPT is built upon a dataset of academic assignments and accompanying metadata [17] and employs a pretrained LLM to compute perplexity scores for student-authored and ChatGPT-generated responses. These scores then assist in establishing a threshold for discerning the origin of a submitted assignment. Given the specificity and contextual nature of academic work, HowkGPT further refines its analysis by defining category-specific thresholds derived from the metadata, enhancing the precision of the detection. This study emphasizes the critical need for effective strategies to uphold academic integrity amidst the growing influence of LLMs and provides an approach to ensuring fair and accurate grading in educational institutions.

cs.CL↗

arXiv2025

Self-Reflection Makes Large Language Models Safer, Less Biased, and Ideologically Neutral

Previous studies proposed that the reasoning capabilities of large language models (LLMs) can be improved through self-reflection, i.e., letting LLMs reflect on their own output to identify and correct mistakes in the initial responses. However, earlier experiments offer mixed results when it comes to the benefits of self-reflection. Furthermore, prior studies on self-reflection are predominantly concerned with the reasoning capabilities of models, ignoring the potential for self-reflection in safety, bias, and ideological leaning. Here, by conducting a series of experiments testing LLM's self-reflection capability in various tasks using a variety of prompts and different LLMs, we make several contributions to the literature. First, we reconcile conflicting findings regarding the benefit of self-reflection, by demonstrating that the outcome of self-reflection is sensitive to prompt wording -- both the original prompt that are used to elicit an initial answer and the subsequent prompt used to self-reflect. Specifically, although self-reflection may improve the reasoning capability of LLMs when the initial response is simple, the technique cannot improve upon the state-of-the-art chain-of-thought (CoT) prompting. Second, we show that self-reflection can lead to safer (75.8\% reduction in toxic responses while preserving 97.8\% non-toxic ones), less biased (77\% reduction in gender biased responses, while preserving 94.3\% unbiased ones), and more ideologically neutral responses (100\% reduction in partisan leaning response, while preserving 87.7\% non-partisan ones). The paper concludes by discussing the implications of our findings on the deployment of large language models. We release our experiments at https://github.com/Michael98Liu/self-reflection.

cs.CL↗

arXiv2025

PixLift: Accelerating Web Browsing via AI Upscaling

Accessing the internet in regions with expensive data plans and limited connectivity poses significant challenges, restricting information access and economic growth. Images, as a major contributor to webpage sizes, exacerbate this issue, despite advances in compression formats like WebP and AVIF. The continued growth of complex and curated web content, coupled with suboptimal optimization practices in many regions, has prevented meaningful reductions in web page sizes. This paper introduces PixLift, a novel solution to reduce webpage sizes by downscaling their images during transmission and leveraging AI models on user devices to upscale them. By trading computational resources for bandwidth, PixLift enables more affordable and inclusive web access. We address key challenges, including the feasibility of scaled image requests on popular websites, the implementation of PixLift as a browser extension, and its impact on user experience. Through the analysis of 71.4k webpages, evaluations of three mainstream upscaling models, and a user study, we demonstrate PixLift's ability to significantly reduce data usage without compromising image quality, fostering a more equitable internet.

cs.PF↗

arXiv2025

A Tale of Three Location Trackers: AirTag, SmartTag, and Tile

Bluetooth Low Energy (BLE) location trackers, or "tags", are popular consumer devices for monitoring personal items. These tags rely on their respective network of companion devices that are capable of detecting their BLE signals and relay location information back to the owner. While manufacturers claim that such crowd-sourced approach yields accurate location tracking, the tags' real-world performance characteristics remain insufficiently understood. To this end, this study presents a comprehensive analysis of three major players in the market: Apple's AirTag, Samsung's SmartTag, and Tile. Our methodology combines controlled experiments -- with a known large distribution of location-reporting devices -- as well as in-the-wild experiments -- with no control on the number and kind of reporting devices encountered, thus emulating real-life use-cases. Leveraging data collection techniques improved from prior research, we recruit 22 volunteers traveling across 29 countries, examining the tags' performance under various environments and conditions. Our findings highlight crucial updates in device behavior since previous studies, with AirTag showing marked improvements in location report frequency. Companion device density emerged as the primary determinant of tag performance, overshadowing technological differences between products. Additionally, we find that post-COVID-19 mobility trends could have contributed to enhanced performance for AirTag and SmartTag. Tile, despite its cross-platform compatibility, exhibited notably lower accuracy, particularly in Asia and Africa, due to limited global adoption. Statistical modeling of spatial errors -- measured as the distance between reported and actual tag locations -- shows log-normal distributions across all tags, highlighting the need for improved location estimation methods to reduce occasional significant inaccuracies.

cs.PF↗

arXiv2025

Non-Western Perspectives on Web Inclusivity: A Study of Accessibility Practices in the Global South

The Global South faces unique challenges in achieving digital inclusion due to a heavy reliance on mobile devices for internet access and the prevalence of slow or unreliable networks. While numerous studies have investigated web accessibility within specific sectors such as education, healthcare, and government services, these efforts have been largely constrained to individual countries or narrow contexts, leaving a critical gap in cross-regional, large-scale analysis. This paper addresses this gap by conducting the first large-scale comparative study of mobile web accessibility across the Global South. In this work, we evaluate 100,000 websites from 10 countries in the Global South to provide a comprehensive understanding of accessibility practices in these regions. Our findings reveal that websites from countries with strict accessibility regulations and enforcement tend to adhere better to Web Content Accessibility Guidelines (WCAG) guidelines. However, accessibility violations impact different disability groups in varying ways. Blind and low-vision individuals in the Global South are disproportionately affected, as only 40% of the evaluated websites meet critical accessibility guidelines. This significant shortfall is largely due to developers frequently neglecting to implement valid alt text for images and ARIA descriptions, which are essential specification mechanisms in the HTML standard for the effective operation of screen readers.

cs.SE↗

arXiv2025

Evaluating the Efficacy of Next.js: A Comparative Analysis with React.js on Performance, SEO, and Global Network Equity

This paper investigates the efficacy of Next.js as a framework addressing the challenges posed by React.js, particularly in performance, SEO, and equitable web accessibility. By constructing identical websites and web applications in both frameworks, we aim to evaluate the frameworks' behavior under diverse network conditions and capabilities. Beyond quantitative metrics like First Contentful Paint (FCP) and Time to Interactive (TTI), we incorporate qualitative user feedback to assess real-world usability. Our motivation stems from bridging the digital divide exacerbated by client-side rendering (CSR) frameworks and validating investments in modern technologies for businesses and institutions. Employing a novel LLM-assisted migration workflow, this paper also demonstrates the ease with which developers can transition from React.js to Next.js. Our results highlight Next.js's promise of better overall performance, without any degradation in user interaction experience, showcasing its potential to mitigate disparities in web accessibility and foster global network equity, thus highlighting Next.js as a compelling framework for the future of an inclusive web.

cs.NI↗

arXiv2025

MAML: Towards a Faster Web in Developing Regions

The web experience in developing regions remains subpar, primarily due to the growing complexity of modern webpages and insufficient optimization by content providers. Users in these regions typically rely on low-end devices and limited bandwidth, which results in a poor user experience as they download and parse webpages bloated with excessive third-party CSS and JavaScript (JS). To address these challenges, we introduce the Mobile Application Markup Language (MAML), a flat layout-based web specification language that reduces computational and data transmission demands, while replacing the excessive bloat from JS with a new scripting language centered on essential (and popular) web functionalities. Last but not least, MAML is backward compatible as it can be transpiled to minimal HTML/JavaScript/CSS and thus work with legacy browsers. We benchmark MAML in terms of page load times and sizes, using a translator which can automatically port any webpage to MAML. When compared to the popular Google AMP, across 100 testing webpages, MAML offers webpage speedups by tens of seconds under challenging network conditions thanks to its significant size reductions. Next, we run a competition involving 25 university students porting 50 of the above webpages to MAML using a web-based editor we developed. This experiment verifies that, with little developer effort, MAML is quite effective in maintaining the visual and functional correctness of the originating webpages.

cs.NI↗

arXiv2024

Advancing Vehicle Plate Recognition: Multitasking Visual Language Models with VehiclePaliGemma

License plate recognition (LPR) involves automated systems that utilize cameras and computer vision to read vehicle license plates. Such plates collected through LPR can then be compared against databases to identify stolen vehicles, uninsured drivers, crime suspects, and more. The LPR system plays a significant role in saving time for institutions such as the police force. In the past, LPR relied heavily on Optical Character Recognition (OCR), which has been widely explored to recognize characters in images. Usually, collected plate images suffer from various limitations, including noise, blurring, weather conditions, and close characters, making the recognition complex. Existing LPR methods still require significant improvement, especially for distorted images. To fill this gap, we propose utilizing visual language models (VLMs) such as OpenAI GPT4o, Google Gemini 1.5, Google PaliGemma (Pathways Language and Image model + Gemma model), Meta Llama 3.2, Anthropic Claude 3.5 Sonnet, LLaVA, NVIDIA VILA, and moondream2 to recognize such unclear plates with close characters. This paper evaluates the VLM's capability to address the aforementioned problems. Additionally, we introduce ``VehiclePaliGemma'', a fine-tuned Open-sourced PaliGemma VLM designed to recognize plates under challenging conditions. We compared our proposed VehiclePaliGemma with state-of-the-art methods and other VLMs using a dataset of Malaysian license plates collected under complex conditions. The results indicate that VehiclePaliGemma achieved superior performance with an accuracy of 87.6\%. Moreover, it is able to predict the car's plate at a speed of 7 frames per second using A100-80GB GPU. Finally, we explored the multitasking capability of VehiclePaliGemma model to accurately identify plates containing multiple cars of various models and colors, with plates positioned and oriented in different directions.

cs.CV↗

arXiv2024

The GAIUS Experience: Powering a Hyperlocal Mobile Web for Communities in Emerging Regions

Despite increasing mobile Internet penetration in developing regions, mobile users continue to experience a poor web experience due to two key factors: (i) lack of locally relevant content; (ii) poor web performance due to complex web pages and poor network conditions. In this paper, we describe our design, implementation and deployment experiences of GAIUS, a mobile content ecosystem that enables efficient creation and dissemination of locally relevant web content into hyperlocal communities in emerging markets. The basic building blocks of GAIUS are a lightweight content edge platform combined with a mobile application that collectively provide a Hyperlocal Web abstraction for mobile users to create and consume locally relevant content and interact with other users via a community abstraction. The GAIUS platform uses MAML, a web specification language that dramatically simplifies web pages to reduce the complexity of Web content within the GAIUS ecosystem, improve page load times and reduce network costs. In this paper, we describe our experiences deploying GAIUS across a large user base in India, Bangladesh and Kenya.

cs.NI↗

arXiv2024

Challenges and Opportunities for Global Cellular Connectivity

Traditional cellular service was designed for global connectivity, but business and logistical constraints led to its fragmentation, with deployments limited to individual countries and regions. Initiatives like Mobile Virtual Network Operators (MVNOs), Mobile Network Aggregators (MNAs), and regulations like ''roam-like-at-home'' have partially restored global service potential, though often at high costs in terms of user bills, application performance, and traffic efficiency. This paper makes two key contributions: first, it surveys the global cellular ecosystem, analyzing the strengths and weaknesses of major players using data from prior research, proprietary datasets, and public sources. Second, it argues that the technology for seamless global service exists in Local Breakout (LBO), a roaming architecture which allows user traffic to be routed directly to the Internet through the visited network, bypassing the home network and/or third-party infrastructures. However, LBO adoption is hindered by issues such as policy enforcement, billing, and Quality of Service (QoS) guarantees, rooted in a lack of trust between operators. The paper concludes by exploring technological advances that could enable LBO, and pave the way for truly global cellular connectivity.

cs.NI↗

arXiv2024

Analyzing political stances on Twitter in the lead-up to the 2024 U.S. election

Social media platforms play a pivotal role in shaping public opinion and amplifying political discourse, particularly during elections. However, the same dynamics that foster democratic engagement can also exacerbate polarization. To better understand these challenges, here, we investigate the ideological positioning of tweets related to the 2024 U.S. Presidential Election. To this end, we analyze 1,235 tweets from key political figures and 63,322 replies, and classify ideological stances into Pro-Democrat, Anti-Republican, Pro-Republican, Anti-Democrat, and Neutral categories. Using a classification pipeline involving three large language models (LLMs)-GPT-4o, Gemini-Pro, and Claude-Opus-and validated by human annotators, we explore how ideological alignment varies between candidates and constituents. We find that Republican candidates author significantly more tweets in criticism of the Democratic party and its candidates than vice versa, but this relationship does not hold for replies to candidate tweets. Furthermore, we highlight shifts in public discourse observed during key political events. By shedding light on the ideological dynamics of online political interactions, these results provide insights for policymakers and platforms seeking to address polarization and foster healthier political dialogue.

cs.SI↗

arXiv2024

Advancing Content Moderation: Evaluating Large Language Models for Detecting Sensitive Content Across Text, Images, and Videos

The widespread dissemination of hate speech, harassment, harmful and sexual content, and violence across websites and media platforms presents substantial challenges and provokes widespread concern among different sectors of society. Governments, educators, and parents are often at odds with media platforms about how to regulate, control, and limit the spread of such content. Technologies for detecting and censoring the media contents are a key solution to addressing these challenges. Techniques from natural language processing and computer vision have been used widely to automatically identify and filter out sensitive content such as offensive languages, violence, nudity, and addiction in both text, images, and videos, enabling platforms to enforce content policies at scale. However, existing methods still have limitations in achieving high detection accuracy with fewer false positives and false negatives. Therefore, more sophisticated algorithms for understanding the context of both text and image may open rooms for improvement in content censorship to build a more efficient censorship system. In this paper, we evaluate existing LLM-based content moderation solutions such as OpenAI moderation model and Llama-Guard3 and study their capabilities to detect sensitive contents. Additionally, we explore recent LLMs such as GPT, Gemini, and Llama in identifying inappropriate contents across media outlets. Various textual and visual datasets like X tweets, Amazon reviews, news articles, human photos, cartoons, sketches, and violence videos have been utilized for evaluation and comparison. The results demonstrate that LLMs outperform traditional techniques by achieving higher accuracy and lower false positive and false negative rates. This highlights the potential to integrate LLMs into websites, social media platforms, and video-sharing services for regulatory and content moderation purposes.

cs.CV↗