arXiv subjects

Mohamed Ali

Publications and source records attributed to Mohamed Ali.

9 recordsLinked to original sources

arXiv2026

GRPO Beyond English: A Large-Scale Study of GRPO in Non-English and Multilingual Settings

Reinforcement Learning with Verifiable Rewards (RLVR), often optimized with Group Relative Policy Optimization (GRPO), has become a central recipe for improving the reasoning capabilities of pretrained language models but current studies remain heavily English-centric. We conduct a large-scale empirical study of multilingual and non-English GRPO across a wide range of base models, training languages, and different reasoning language rewards. We find that training to reason in the native language often leaves only a small gap to training for English reasoning. We further observe strong crosslingual transfer: training in one language often improves performance in many others. However, specific trends are highly model- and language-dependent. In some cases, training in a particular language induces severe regressions on out-of-domain capabilities in other languages. Our analysis shows that RLVR beyond English can provide broad crosslingual gains, but also requires broad evaluation to detect language-specific regressions.

cs.CL↗

arXiv2026

mAceReason-Math: A Dataset of High-Quality Multilingual Math Problems Ready For RLVR

Reinforcement Learning with Verifiable Rewards (RLVR) has been successfully applied to significantly boost the capabilities of pretrained large language models, especially in the math and logic problem domains. However, current research and available training datasets remain English-centric. While multilingual training data and benchmarks have been created in the past, they were not created with RLVR and current model capability in mind, and their level of difficulty is often too low to provide appropriate training signals for current models. To address this gap, we provide mAceReason-Math, a dataset of high-quality translations of challenging math problems sourced from a corpus specifically curated for RLVR (AceReason-Math). We further take specific care to clean and improve our translations, resulting in a coverage of 14 languages with more than 10,000 samples per language. We release the dataset to facilitate multilingual RLVR research and benchmarking in the research community.

cs.CL↗

arXiv2026

Multilingual Reasoning Gym: Multilingual Scaling of Procedural Reasoning Environments

We present the Multilingual Reasoning Gym, an extension of Reasoning Gym (Stojanovski et al., 2025), that procedurally generates verifiable reasoning problems across 14 languages. We translate templates for 94 tasks with native-speaker validation in 10 languages and targeted code or template adaptations to ensure linguistic naturalness. The Multilingual Reasoning Gym preserves the core benefits of the procedural generation approach used in the original Reasoning Gym, such as virtually unlimited problem instance generation and adjustable difficulty, and remains directly usable for Reinforcement Learning from Verifiable Rewards and evaluation settings. Problems in the Multilingual Reasoning Gym are parallel across languages, enabling crosslingually parallel data generation at massive scale due to the procedural nature of the environments. We release our implementation to support research into multilingual reasoning models.

cs.CL↗

arXiv2025

FASL-Seg: Anatomy and Tool Segmentation of Surgical Scenes

The growing popularity of robotic minimally invasive surgeries has made deep learning-based surgical training a key area of research. A thorough understanding of the surgical scene components is crucial, which semantic segmentation models can help achieve. However, most existing work focuses on surgical tools and overlooks anatomical objects. Additionally, current state-of-the-art (SOTA) models struggle to balance capturing high-level contextual features and low-level edge features. We propose a Feature-Adaptive Spatial Localization model (FASL-Seg), designed to capture features at multiple levels of detail through two distinct processing streams, namely a Low-Level Feature Projection (LLFP) and a High-Level Feature Projection (HLFP) stream, for varying feature resolutions - enabling precise segmentation of anatomy and surgical instruments. We evaluated FASL-Seg on surgical segmentation benchmark datasets EndoVis18 and EndoVis17 on three use cases. The FASL-Seg model achieves a mean Intersection over Union (mIoU) of 72.71% on parts and anatomy segmentation in EndoVis18, improving on SOTA by 5%. It further achieves a mIoU of 85.61% and 72.78% in EndoVis18 and EndoVis17 tool type segmentation, respectively, outperforming SOTA overall performance, with comparable per-class SOTA results in both datasets and consistent performance in various classes for anatomy and instruments, demonstrating the effectiveness of distinct processing streams for varying feature resolutions.

eess.IV↗

arXiv2025

Scalability, Availability, Reproducibility and Extensibility in Islamic Database Systems

With the widespread of software systems and applications that serve the Islamic knowledge domain, several concerns arise. Authenticity and accuracy of the databases that back up these systems are questionable. With the excitement that some software developers and amateur researchers may have, false statements and incorrect claims may be made around numerical signs or miracles in the Quran. Reproducibility of these claims may not be addressed by the people making such claims. Moreover, with the increase in the number of users, scalability and availability of these systems become a concern. In addition to all these concerns, extensibility is also another major issue. Properly designed systems can be extensible, reusable and built on top of one another, instead of each system being built from scratch every time a new framework is developed. In this paper, we introduce the QuranResearch.Org system and its vision for scalability, availability, reproducibility and extensibility to serve Islamic database systems.

cs.DB↗

arXiv2024

Symmetry Breaking in Chemical Systems: Engineering Complexity through Self-Organization and Marangoni Flows

Far from equilibrium, chemical and biological systems can form complex patterns and waves through reaction-diffusion coupling. Fluid motion often interferes with these self-organized concentration patterns. In this study, we investigate the influence of Marangoni-driven flows inside a thin layer of fluid ascending the outer surfaces of hydrophilic obstacles on the spatio-temporal dynamics of chemical waves in the modified Belousov-Zhabotinsky reaction. Our observations reveal that circular waves originate nearly simultaneously at the obstacles and propagate outward. In a covered setup, where evaporation is minimal, the wavefronts maintain their circular shape. However, in an uncovered setup with significant evaporation and resulting Marangoni flows, the interplay between surface tension-driven Marangoni flows and gravity destabilizes the wavefronts, creating distinctive flower-like patterns around the obstacles. Our analysis shows that here solutal Marangoni forces are more relevant than thermal ones. Our experiments further show that the number of petals formed increases linearly with the obstacle's diameter, though a minimum diameter is required for these instabilities to appear. These findings demonstrate the potential to 'engineer' specific wave patterns, offering a method to control and direct reaction dynamics. This capability is especially important for developing microfluidic devices requiring precise control over chemical wave propagation.

physics.flu-dyn↗

arXiv2022

Optimizing Bilingual Neural Transducer with Synthetic Code-switching Text Generation

Code-switching describes the practice of using more than one language in the same sentence. In this study, we investigate how to optimize a neural transducer based bilingual automatic speech recognition (ASR) model for code-switching speech. Focusing on the scenario where the ASR model is trained without supervised code-switching data, we found that semi-supervised training and synthetic code-switched data can improve the bilingual ASR system on code-switching speech. We analyze how each of the neural transducer's encoders contributes towards code-switching performance by measuring encoder-specific recall values, and evaluate our English/Mandarin system on the ASCEND data set. Our final system achieves 25% mixed error rate (MER) on the ASCEND English/Mandarin code-switching test set -- reducing the MER by 2.1% absolute compared to the previous literature -- while maintaining good accuracy on the monolingual test sets.

cs.SD↗

arXiv2016

Towards Policy Enforcement Point as a Service (PEPS)

In this paper, we coin the term Policy Enforcement as a Service (PEPS), which enables the provision of innovative inter-layer and inter-domain Access Control. We leverage the architecture of Software-Defined-Network (SDN) to introduce a common network-level enforcement point, which is made available to a range of access control systems. With our PEPS model, it is possible to have a `defense in depth' protection model and drop unsuccessful access requests before engaging the data provider (e.g. a database system). Moreover, the current implementation of access control within the `trusted' perimeter of an organization is no longer a restriction so that the potential for novel, distributed and cooperative security services can be realized. We conduct an analysis of the security requirements and technical challenges for implementing Policy Enforcement as a Service. To illustrate the benefits of our proposal in practice, we include a report on our prototype PEPS-enabled location-based access control.

cs.CR↗

arXiv2006

Consistent Streaming Through Time: A Vision for Event Stream Processing

Event processing will play an increasingly important role in constructing enterprise applications that can immediately react to business critical events. Various technologies have been proposed in recent years, such as event processing, data streams and asynchronous messaging (e.g. pub/sub). We believe these technologies share a common processing model and differ only in target workload, including query language features and consistency requirements. We argue that integrating these technologies is the next step in a natural progression. In this paper, we present an overview and discuss the foundations of CEDR, an event streaming system that embraces a temporal stream model to unify and further enrich query language features, handle imperfections in event delivery and define correctness guarantees. We describe specific contributions made so far and outline next steps in developing the CEDR system.

cs.DB↗