arXiv subjects

Mohamed Mahmoud

Publications and source records attributed to Mohamed Mahmoud.

At least 19 recordsLinked to original sources

arXiv2026

RAVEN: Retrieval-Augmented Vulnerability Exploration Network for Memory Corruption Analysis in User Code and Binary Programs

Large Language Models (LLMs) have demonstrated remarkable capabilities across various cybersecurity tasks, including vulnerability classification, detection, and patching. However, their potential in automated vulnerability report documentation and analysis remains underexplored. We present RAVEN (Retrieval Augmented Vulnerability Exploration Network), a framework leveraging LLM agents and Retrieval Augmented Generation (RAG) to synthesize comprehensive vulnerability analysis reports. Given vulnerable source code, RAVEN generates reports following the Google Project Zero Root Cause Analysis template. The framework uses four modules: an Explorer agent for vulnerability identification, a RAG engine retrieving relevant knowledge from curated databases including Google Project Zero reports and CWE entries, an Analyst agent for impact and exploitation assessment, and a Reporter agent for structured report generation. To ensure quality, RAVEN includes a task specific LLM Judge evaluating reports across structural integrity, ground truth alignment, code reasoning quality, and remediation quality. We evaluate RAVEN on 105 vulnerable code samples covering 15 CWE types from the NIST-SARD dataset. Results show an average quality score of 54.21%, supporting the effectiveness of our approach for automated vulnerability documentation.

cs.CR↗

arXiv2026

MARVEL: Multimodal Adaptive Reasoning-intensiVe Expand-rerank and retrievaL

Multimodal retrieval over text corpora remains a fundamental challenge: the best vision-language encoder achieves only 27.6 nDCG@10 on MM-BRIGHT, a reasoning-intensive multimodal retrieval benchmark, underperforming strong text-only systems. We argue that effective multimodal retrieval requires three tightly integrated capabilities that existing approaches address only in isolation: expanding the query's latent intent, retrieving with a model trained for complex reasoning, and reranking via explicit step-by-step reasoning over candidates. We introduce \textbf{MARVEL} (\textbf{M}ultimodal \textbf{A}daptive \textbf{R}easoning-intensi\textbf{V}e \textbf{E}xpand-rerank and retrieva\textbf{L}), a unified pipeline that combines LLM-driven query expansion, \textbf{MARVEL-Retriever} -- a reasoning-enhanced dense retriever fine-tuned for complex multimodal queries -- and GPT-4o-based chain-of-thought reranking with optional multi-pass reciprocal rank fusion. Evaluated on MM-BRIGHT across 29 technical domains, MARVEL achieves \textbf{37.9} nDCG@10, surpassing the best multimodal encoder by \textbf{+10.3 points} and outperforming all single-stage baselines in 27 of 29 domains and matching or approaching the best baseline in the remaining two highly-specialized domains (Crypto, Quantum Computing), demonstrating that reasoning-intensive multimodal retrieval is best addressed through a unified expand-retrieve-rerank framework. https://github.com/mm-bright/multimodal-reasoning-retrieval

cs.IR↗

arXiv2026

BRIDGE: Multimodal-to-Text Retrieval via Reinforcement-Learned Query Alignment

Multimodal retrieval systems struggle to resolve image-text queries against text-only corpora: the best vision-language encoder achieves only 27.6 nDCG@10 on MM-BRIGHT, underperforming strong text-only retrievers. We argue the bottleneck is not the retriever but the query -- raw multimodal queries entangle visual descriptions, conversational noise, and retrieval intent in ways that systematically degrade embedding similarity. We present \textbf{BRIDGE}, a two-component system that resolves this mismatch without multimodal encoders. \textbf{FORGE} (\textbf{F}ocused Retrieval Query Generato\textbf{r}) is a query alignment model trained via reinforcement learning, which distills noisy multimodal queries into compact, retrieval-optimized search strings. \textbf{LENS} (\textbf{L}anguage-\textbf{E}nhanced \textbf{N}eural \textbf{S}earch) is a reasoning-enhanced dense retriever fine-tuned on reasoning-intensive retrieval data to handle the intent-rich queries FORGE produces. Evaluated on MM-BRIGHT (2,803 queries, 29 domains), BRIDGE achieves \textbf{29.7} nDCG@10, surpassing all multimodal encoder baselines including Nomic-Vision (27.6). When FORGE is applied as a plug-and-play aligner on top of Nomic-Vision, the combined system reaches \textbf{33.3} nDCG@10 -- exceeding the best text-only retriever (32.2) -- demonstrating that \textit{query alignment} is the key bottleneck in multimodal-to-text retrieval. https://github.com/mm-bright/multimodal-reasoning-retrieval

cs.IR↗

arXiv2026

HIVE: Query, Hypothesize, Verify An LLM Framework for Multimodal Reasoning-Intensive Retrieval

Multimodal retrieval models fail on reasoning-intensive queries where images (diagrams, charts, screenshots) must be deeply integrated with text to identify relevant documents -- the best multimodal model achieves only 27.6 nDCG@10 on MM-BRIGHT, underperforming even strong text-only retrievers (32.2). We introduce \textbf{HIVE} (\textbf{H}ypothesis-driven \textbf{I}terative \textbf{V}isual \textbf{E}vidence Retrieval), a plug-and-play framework that injects explicit visual-text reasoning into a retriever via LLMs. HIVE operates in four stages: (1) initial retrieval over the corpus, (2) LLM-based compensatory query synthesis that explicitly articulates visual and logical gaps observed in top-$k$ candidates, (3) secondary retrieval with the refined query, and (4) LLM verification and reranking over the union of candidates. Evaluated on the multimodal-to-text track of MM-BRIGHT (2,803 real-world queries across 29 technical domains), HIVE achieves a new state-of-the-art aggregated nDCG@10 of \textbf{41.7} -- a \textbf{+9.5} point gain over the best text-only model (DiVeR: 32.2) and \textbf{+14.1} over the best multimodal model (Nomic-Vision: 27.6), where our reasoning-enhanced base retriever contributes 33.2 and the HIVE framework adds a further \textbf{+8.5} points -- with particularly strong results in visually demanding domains (Gaming: 68.2, Chemistry: 42.5, Sustainability: 49.4). Compatible with both standard and reasoning-enhanced retrievers, HIVE demonstrates that LLM-mediated visual hypothesis generation and verification can substantially close the multimodal reasoning gap in retrieval. https://github.com/mm-bright/multimodal-reasoning-retrieval

cs.IR↗

arXiv2026

Representation-Level Adversarial Regularization for Clinically Aligned Multitask Thyroid Ultrasound Assessment

Thyroid ultrasound is the first-line exam for assessing thyroid nodules and determining whether biopsy is warranted. In routine reporting, radiologists produce two coupled outputs: a nodule contour for measurement and a TI-RADS risk category based on sonographic criteria. Yet both contouring style and risk grading vary across readers, creating inconsistent supervision that can degrade standard learning pipelines. In this paper, we address this workflow with a clinically guided multitask framework that jointly predicts the nodule mask and TI-RADS category within a single model. To ground risk prediction in clinically meaningful evidence, we guide the classification embedding using a compact TI-RADS aligned radiomics target during training, while preserving complementary deep features for discriminative performance. However, under annotator variability, naive multitask optimization often fails not because the tasks are unrelated, but because their gradients compete within the shared representation. To make this competition explicit and controllable, we introduce RLAR, a representation-level adversarial gradient regularizer. Rather than performing parameter-level gradient surgery, RLAR uses each task's normalized adversarial direction in latent space as a geometric probe of task sensitivity and penalizes excessive angular alignment between task-specific adversarial directions. On a public TI-RADS dataset, our clinically guided multitask model with RLAR consistently improves risk stratification while maintaining segmentation quality compared to single-task training and conventional multitask baselines. Code and pretrained models will be released.

cs.CV↗

arXiv2026

MM-BRIGHT: A Multi-Task Multimodal Benchmark for Reasoning-Intensive Retrieval

Existing retrieval benchmarks primarily consist of text-based queries where keyword or semantic matching is usually sufficient. Many real-world queries contain multimodal elements, particularly, images such as diagrams, charts, and screenshots that require intensive reasoning to identify relevant documents. To address this gap, we introduce MM-BRIGHT, the first multimodal benchmark for reasoning-intensive retrieval. Our dataset consists of 2,803 real-world queries spanning 29 diverse technical domains, with four tasks of increasing complexity: text-to-text, multimodal-to-text, multimodal-to-image, and multimodal-to-multimodal retrieval. Extensive evaluation reveals that state-of-the-art models struggle across all tasks: BM25 achieves only 8.5 nDCG@10 on text-only retrieval, while the best multimodal model Nomic-Vision reaches just 27.6 nDCG@10 on multimodal-to-text retrieval actually underperforming the best text-only model (DiVeR: 32.2). These results highlight substantial headroom and position MM-BRIGHT as a testbed for next-generation retrieval models that better integrate visual reasoning. Our code and data are available at https://github.com/mm-bright/MM-BRIGHT. See also our official website: https://mm-bright.github.io/.

cs.IR↗

arXiv2025

Optimisation of Power Modulation for Hall-Héroult Cells: Process Operability and Constraints as Virtual Energy Storage

Aluminium is manufactured through the Hall-Héroult process, which is very energy intensive. Power modulation, as an industrial-scale demand-side power management approach, allows aluminium smelters to operate with variable power consumption rates and as such be powered by renewable energy sources. In this way, aluminium smelting cells can be used as a large virtual energy storage to balance power demand-supply and stabilise electrical grids. This paper studies the potential optimal power modulation operating conditions, including time-varying line current and anode-cathode distance (ACD) profiles to maximise the aluminium reduction cell profitability subject to constraints on the cell thermal balance. To deal with the complex cell dynamics which are spatially distributed and multi-timescale, a novel optimisation approach that utilises both reduced-order and detailed models is developed. The results yield insight into the optimal line current and ACD profiles for different power modulation scenarios including the time of use electricity tariff and spot price. These results can form the foundation for further studies into online control policies of aluminium reduction cells.

eess.SY↗

arXiv2025

ReceiptSense: Beyond Traditional OCR -- A Dataset for Receipt Understanding

Multilingual OCR and information extraction from receipts remains challenging, particularly for complex scripts like Arabic. We introduce \dataset, a comprehensive dataset designed for Arabic-English receipt understanding comprising 20,000 annotated receipts from diverse retail settings, 30,000 OCR-annotated images, and 10,000 item-level annotations, and a new Receipt QA subset with 1265 receipt images paired with 40 question-answer pairs each to support LLM evaluation for receipt understanding. The dataset captures merchant names, item descriptions, prices, receipt numbers, and dates to support object detection, OCR, and information extraction tasks. We establish baseline performance using traditional methods (Tesseract OCR) and advanced neural networks, demonstrating the dataset's effectiveness for processing complex, noisy real-world receipt layouts. Our publicly accessible dataset advances automated multilingual document processing research (see https://github.com/Update-For-Integrated-Business-AI/CORU ).

cs.CV↗

arXiv2025

Foundation Models as Class-Incremental Learners for Dermatological Image Classification

Class-Incremental Learning (CIL) aims to learn new classes over time without forgetting previously acquired knowledge. The emergence of foundation models (FM) pretrained on large datasets presents new opportunities for CIL by offering rich, transferable representations. However, their potential for enabling incremental learning in dermatology remains largely unexplored. In this paper, we systematically evaluate frozen FMs pretrained on large-scale skin lesion datasets for CIL in dermatological disease classification. We propose a simple yet effective approach where the backbone remains frozen, and a lightweight MLP is trained incrementally for each task. This setup achieves state-of-the-art performance without forgetting, outperforming regularization, replay, and architecture based methods. To further explore the capabilities of frozen FMs, we examine zero training scenarios using nearest mean classifiers with prototypes derived from their embeddings. Through extensive ablation studies, we demonstrate that this prototype based variant can also achieve competitive results. Our findings highlight the strength of frozen FMs for continual learning in dermatology and support their broader adoption in real world medical applications. Our code and datasets are available here.

cs.CV↗

arXiv2025

PDLRecover: Privacy-preserving Decentralized Model Recovery with Machine Unlearning

Decentralized learning is vulnerable to poison attacks, where malicious clients manipulate local updates to degrade global model performance. Existing defenses mainly detect and filter malicious models, aiming to prevent a limited number of attackers from corrupting the global model. However, restoring an already compromised global model remains a challenge. A direct approach is to remove malicious clients and retrain the model using only the benign clients. Yet, retraining is time-consuming, computationally expensive, and may compromise model consistency and privacy. We propose PDLRecover, a novel method to recover a poisoned global model efficiently by leveraging historical model information while preserving privacy. The main challenge lies in protecting shared historical models while enabling parameter estimation for model recovery. By exploiting the linearity of approximate Hessian matrix computation, we apply secret sharing to protect historical updates, ensuring local models are not leaked during transmission or reconstruction. PDLRecover introduces client-side preparation, periodic recovery updates, and a final exact update to ensure robustness and convergence of the recovered model. Periodic updates maintain accurate curvature information, and the final step ensures high-quality convergence. Experiments show that the recovered global model achieves performance comparable to a fully retrained model but with significantly reduced computation and time cost. Moreover, PDLRecover effectively prevents leakage of local model parameters, ensuring both accuracy and privacy in recovery.

cs.CR↗

arXiv2024

A Comprehensive Survey of Masked Faces: Recognition, Detection, and Unmasking

Masked face recognition (MFR) has emerged as a critical domain in biometric identification, especially by the global COVID-19 pandemic, which introduced widespread face masks. This survey paper presents a comprehensive analysis of the challenges and advancements in recognising and detecting individuals with masked faces, which has seen innovative shifts due to the necessity of adapting to new societal norms. Advanced through deep learning techniques, MFR, along with Face Mask Recognition (FMR) and Face Unmasking (FU), represent significant areas of focus. These methods address unique challenges posed by obscured facial features, from fully to partially covered faces. Our comprehensive review delves into the various deep learning-based methodologies developed for MFR, FMR, and FU, highlighting their distinctive challenges and the solutions proposed to overcome them. Additionally, we explore benchmark datasets and evaluation metrics specifically tailored for assessing performance in MFR research. The survey also discusses the substantial obstacles still facing researchers in this field and proposes future directions for the ongoing development of more robust and effective masked face recognition systems. This paper serves as an invaluable resource for researchers and practitioners, offering insights into the evolving landscape of face recognition technologies in the face of global health crises and beyond.

cs.CV↗

arXiv2024

ArabicaQA: A Comprehensive Dataset for Arabic Question Answering

In this paper, we address the significant gap in Arabic natural language processing (NLP) resources by introducing ArabicaQA, the first large-scale dataset for machine reading comprehension and open-domain question answering in Arabic. This comprehensive dataset, consisting of 89,095 answerable and 3,701 unanswerable questions created by crowdworkers to look similar to answerable ones, along with additional labels of open-domain questions marks a crucial advancement in Arabic NLP resources. We also present AraDPR, the first dense passage retrieval model trained on the Arabic Wikipedia corpus, specifically designed to tackle the unique challenges of Arabic text retrieval. Furthermore, our study includes extensive benchmarking of large language models (LLMs) for Arabic question answering, critically evaluating their performance in the Arabic language context. In conclusion, ArabicaQA, AraDPR, and the benchmarking of LLMs in Arabic question answering offer significant advancements in the field of Arabic NLP. The dataset and code are publicly accessible for further research https://github.com/DataScienceUIBK/ArabicaQA.

cs.CL↗

arXiv2023

Advancements and Challenges in Arabic Optical Character Recognition: A Comprehensive Survey

Optical character recognition (OCR) is a vital process that involves the extraction of handwritten or printed text from scanned or printed images, converting it into a format that can be understood and processed by machines. This enables further data processing activities such as searching and editing. The automatic extraction of text through OCR plays a crucial role in digitizing documents, enhancing productivity, improving accessibility, and preserving historical records. This paper seeks to offer an exhaustive review of contemporary applications, methodologies, and challenges associated with Arabic Optical Character Recognition (OCR). A thorough analysis is conducted on prevailing techniques utilized throughout the OCR process, with a dedicated effort to discern the most efficacious approaches that demonstrate enhanced outcomes. To ensure a thorough evaluation, a meticulous keyword-search methodology is adopted, encompassing a comprehensive analysis of articles relevant to Arabic OCR, including both backward and forward citation reviews. In addition to presenting cutting-edge techniques and methods, this paper critically identifies research gaps within the realm of Arabic OCR. By highlighting these gaps, we shed light on potential areas for future exploration and development, thereby guiding researchers toward promising avenues in the field of Arabic OCR. The outcomes of this study provide valuable insights for researchers, practitioners, and stakeholders involved in Arabic OCR, ultimately fostering advancements in the field and facilitating the creation of more accurate and efficient OCR systems for the Arabic language.

cs.CV↗

arXiv2022

Deep learning for table detection and structure recognition: A survey

Tables are everywhere, from scientific journals, papers, websites, and newspapers all the way to items we buy at the supermarket. Detecting them is thus of utmost importance to automatically understanding the content of a document. The performance of table detection has substantially increased thanks to the rapid development of deep learning networks. The goals of this survey are to provide a profound comprehension of the major developments in the field of Table Detection, offer insight into the different methodologies, and provide a systematic taxonomy of the different approaches. Furthermore, we provide an analysis of both classic and new applications in the field. Lastly, the datasets and source code of the existing models are organized to provide the reader with a compass on this vast literature. Finally, we go over the architecture of utilizing various object detection and table structure recognition methods to create an effective and efficient system, as well as a set of development trends to keep up with state-of-the-art algorithms and future research. We have also set up a public GitHub repository where we will be updating the most recent publications, open data, and source code. The GitHub repository is available at https://github.com/abdoelsayed2016/table-detection-structure-recognition.

cs.CV↗

arXiv2021

Efficient and Privacy-Preserving Infection Control System for Covid-19-Like Pandemics using Blockchain

Contact tracing is a very effective way to control the COVID-19-like pandemics. It aims to identify individuals who closely contacted an infected person during the incubation period of the virus and notify them to quarantine. However, the existing systems suffer from privacy, security, and efficiency issues. To address these limitations, in this paper, we propose an efficient and privacy-preserving Blockchain-based infection control system. Instead of depending on a single authority to run the system, a group of health authorities, that form a consortium Blockchain, run our system. Using Blockchain technology not only secures our system against single point of failure and denial of service attacks, but also brings transparency because all transactions can be validated by different parties. Although contact tracing is important, it is not enough to effectively control an infection. Thus, unlike most of the existing systems that focus only on contact tracing, our system consists of three integrated subsystems, including contact tracing, public places access control, and safe-places recommendation. The access control subsystem prevents infected people from visiting public places to prevent spreading the virus, and the recommendation subsystem categorizes zones based on the infection level so that people can avoid visiting contaminated zones. Our analysis demonstrates that our system is secure and preserves the privacy of the users against identification, social graph disclosure, and tracking attacks, while thwarting false reporting (or panic) attacks. Moreover, our extensive performance evaluations demonstrate the scalability of our system (which is desirable in pandemics) due to its low communication, computation, and storage overheads.

cs.CR↗

arXiv2021

Privacy-Preserving Smart Parking System Using Blockchain and Private Information Retrieval

Searching for available parking spaces is a major problem for drivers especially in big crowded cities, causing traffic congestion and air pollution, and wasting drivers' time. Smart parking systems are a novel solution to enable drivers to have real-time parking information for pre-booking. However, current smart parking requires drivers to disclose their private information, such as desired destinations. Moreover, the existing schemes are centralized and vulnerable to the bottleneck of the single point of failure and data breaches. In this paper, we propose a distributed privacy-preserving smart parking system using blockchain. A consortium blockchain created by different parking lot owners to ensure security, transparency, and availability is proposed to store their parking offers on the blockchain. To preserve drivers' location privacy, we adopt a private information retrieval (PIR) technique to enable drivers to retrieve parking offers from blockchain nodes privately, without revealing which parking offers are retrieved. Furthermore, a short randomizable signature is used to enable drivers to reserve available parking slots in an anonymous manner. Besides, we introduce an anonymous payment system that cannot link drivers' to specific parking locations. Finally, our performance evaluations demonstrate that the proposed scheme can preserve drivers' privacy with low communication and computation overhead.

cs.CR↗

arXiv2020

Detection of False-Reading Attacks in the AMI Net-Metering System

In smart grid, malicious customers may compromise their smart meters (SMs) to report false readings to achieve financial gains illegally. Reporting false readings not only causes hefty financial losses to the utility but may also degrade the grid performance because the reported readings are used for energy management. This paper is the first work that investigates this problem in the net-metering system, in which one SM is used to report the difference between the power consumed and the power generated. First, we prepare a benign dataset for the net-metering system by processing a real power consumption and generation dataset. Then, we propose a new set of attacks tailored for the net-metering system to create malicious dataset. After that, we analyze the data and we found time correlations between the net meter readings and correlations between the readings and relevant data obtained from trustworthy sources such as the solar irradiance and temperature. Based on the data analysis, we propose a general multi-data-source deep hybrid learning-based detector to identify the false-reading attacks. Our detector is trained on net meter readings of all customers besides data from the trustworthy sources to enhance the detector performance by learning the correlations between them. The rationale here is that although an attacker can report false readings, he cannot manipulate the solar irradiance and temperature values because they are beyond his control. Extensive experiments have been conducted, and the results indicate that our detector can identify the false-reading attacks with high detection rate and low false alarm.

cs.CR↗

arXiv2020

Privacy-Preserving and Efficient Data Collection Scheme for AMI Networks Using Deep Learning

In advanced metering infrastructure (AMI), smart meters (SMs), which are installed at the consumer side, send fine-grained power consumption readings periodically to the electricity utility for load monitoring and energy management. Change and transmit (CAT) is an efficient approach to collect these readings, where the readings are not transmitted when there is no enough change in consumption. However, this approach causes a privacy problem that is by analyzing the transmission pattern of an SM, sensitive information on the house dwellers can be inferred. For instance, since the transmission pattern is distinguishable when dwellers are on travel, attackers may analyze the pattern to launch a presence-privacy attack (PPA) to infer whether the dwellers are absent from home. In this paper, we propose a scheme, called "STDL", for efficient collection of power consumption readings in AMI networks while preserving the consumers' privacy by sending spoofing transmissions (redundant real readings) using a deep-learning approach. We first use a clustering technique and real power consumption readings to create a dataset for transmission patterns using the CAT approach. Then, we train an attacker model using deep-learning, and our evaluations indicate that the success rate of the attacker is about 91%. Finally, we train a deep-learning-based defense model to send spoofing transmissions efficiently to thwart the PPA. Extensive evaluations are conducted, and the results indicate that our scheme can reduce the attacker's success rate, to 13.52% in case he knows the defense model and to 3.15% in case he does not know the model, while still achieving high efficiency in terms of the number of readings that should be transmitted. Our measurements indicate that the proposed scheme can reduce the number of readings that should be transmitted by about 41% compared to continuously transmitting readings.

cs.CR↗