Fethi Khaled, a mechanical engineering Ph.D. student at KAUST's Clean Combustion Research Center (CCRC), is researching fuel combustion with a focus on cleaner, safer, and more efficient energy sources. His work in the Chemical Kinetics and Laser Sensors Laboratory under Professor Aamir Farooq involves studying the science of combustion and different energy sources like fossil and solar energy. Khaled aims to contribute to inventing new combustion engine modes that are more efficient and produce less or zero pollutants. Why it matters: This research aligns with Saudi Arabia's broader goals of promoting sustainable energy solutions and reducing reliance on traditional fossil fuels, contributing to environmental sustainability and economic diversification.
Ahmed Sultan Salem, a visiting associate professor of electrical engineering, received the 2017 KAUST Distinguished Teaching Award. Salem was one of six finalists nominated for the inaugural award and has been with KAUST since 2011. He teaches a range of EE and applied mathematics courses and his research interests include energy harvesting and cognitive radio technology. Why it matters: Recognizing teaching excellence can help incentivize high-quality education and mentorship in technical fields crucial for advancing Saudi Arabia's research and development goals.
KAUST researchers in the Sensors Lab are developing neuromorphic circuits for vision sensors, drawing inspiration from the human eye. They created flexible photoreceptors using hybrid perovskite materials, with capacitance tunable by light stimulation, mimicking the human retina. The team collaborates with experts in image characterization and brain pattern recognition to connect the 'eye' to the 'brain' for object identification. Why it matters: This biomimetic approach promises advancements in AI, machine learning, and smart city development within the region.
The paper introduces SalamahBench, a new benchmark for evaluating the safety of Arabic Language Models (ALMs). The benchmark comprises 8,170 prompts across 12 categories aligned with the MLCommons Safety Hazard Taxonomy. Five state-of-the-art ALMs, including Fanar 1 and 2, ALLaM 2, Falcon H1R, and Jais 2, were evaluated using the benchmark. Why it matters: The benchmark enables standardized, category-aware safety evaluation, highlighting the necessity of specialized safeguard mechanisms for robust harm mitigation in ALMs.
Adel Bibi, a KAUST alumnus and researcher at the University of Oxford, presented his research on AI safety, covering robustness, alignment, and fairness of LLMs. The research addresses challenges in AI systems, alignment issues, and fairness across languages in common tokenizers. Bibi's work includes instruction prefix tuning and its theoretical limitations towards alignment. Why it matters: This research from a leading researcher highlights the importance of addressing safety concerns in LLMs, particularly regarding alignment and fairness in the Arabic language.
KAUST Ph.D. student Khalil Moussi won two awards at the IEEE International Conference on Nano/Micro Engineered and Molecular Systems for his research on a miniaturized drug delivery system. The system, developed in collaboration with KAIMRC, uses 3D printing and wireless power to deliver drugs for coronary artery disease treatment. The device features an electrochemical micro-pump, a 3D printed reservoir with microneedles, and a wireless powering unit, allowing customization for various in vivo drug delivery applications. Why it matters: This recognition highlights KAUST's contributions to biomedical engineering and its potential to develop innovative solutions for critical healthcare challenges in the region and beyond.
This paper introduces AraLLaMA, a new Arabic large language model (LLM) trained using a progressive vocabulary expansion method inspired by second language acquisition. The model utilizes a modified byte-pair encoding (BPE) algorithm to dynamically extend the Arabic subwords in its vocabulary during training, balancing the out-of-vocabulary (OOV) ratio. Experiments show AraLLaMA achieves performance comparable to existing Arabic LLMs on various benchmarks, and all models, data, and code will be open-sourced. Why it matters: This work addresses the need for more accessible and performant Arabic LLMs, contributing to democratization of AI in the Arab world.