Two professors from the University of California, Merced, are set to explore artificial intelligence in collaboration with Amazon after receiving the tech giant’s prestigious Research Awards. Dong Li and Xiaoyi Lu were announced as recipients in a recent announcement from Amazon, which awarded a total of 63 grants to researchers across 41 universities in eight countries this cycle.
The Amazon Research Awards provide unrestricted funding and Amazon Web Services Promotional Credits to academic researchers focused on various disciplines. Both Li and Lu’s projects will leverage AWS Trainium, a chip designed for high-performance deep learning training of generative AI models. Their work aims to push the boundaries of AI capabilities and improve the efficiency and accessibility of advanced AI technologies.
Li’s project, titled “Efficient Sparse Training with Adaptive Expert Parallelism on AWS Trainium,” focuses on enhancing the learning speed and power efficiency of multiple computers working together. This initiative aims to develop methods that allow for quicker training processes, which can significantly impact energy consumption in AI research.
In parallel, Lu’s project, “Accelerating Large Language and Reasoning Model Workloads with AWS Trainium,” seeks to improve the speed of computational processes involved in training state-of-the-art AI models, such as those utilized by OpenAI’s GPT and Google’s Gemini. Lu’s research will emphasize the efficiency of memory systems and communication to enable high-performance training of AI models.
Amazon evaluates proposals based on their scientific quality and potential societal impact. The company encourages the publication of research findings, presentations at Amazon offices worldwide, and the release of related code under open-source licenses. Award recipients have access to over 700 Amazon public datasets, as well as AWS AI/ML services and tools via their Promotional Credits. Furthermore, they are assigned an Amazon research contact who provides consultation and guidance while also facilitating opportunities to engage in Amazon events and training sessions.
Yida Wang, a principal applied scientist at AWS AI, highlighted the challenges faced by academic AI researchers, emphasizing the need for access to powerful yet affordable infrastructure. “Advancing machine learning research and educating the next generation requires access to cutting-edge infrastructure that’s both powerful and affordable,” he stated. The Build on Trainium program aims to address this barrier, reflecting AWS’s commitment to democratizing AI research through collaborative partnerships with academia.
Both Li and Lu expressed enthusiasm regarding the opportunities that the grants offer. Lu explained that his project aims to “push the performance boundaries of modern AI systems and make advanced AI capabilities more efficient and accessible.” Li echoed this sentiment, noting that the awards recognize the high-quality research in high-performance computing taking place at UC Merced.
Li further elaborated on the benefits the award brings to students, mentioning that it will provide access to state-of-the-art hardware for AI research, along with opportunities for interesting projects and internships. He emphasized the importance of connections between AWS and UC Merced for future collaborations in techniques, education, and business.
As the landscape of artificial intelligence continues to evolve, initiatives like these not only foster innovation in technological research but also enhance educational opportunities for students, potentially shaping the next generation of AI leaders. The collaboration between UC Merced and Amazon underscores a growing trend of partnerships aimed at advancing our understanding and capabilities in AI, promising to have lasting implications for both academia and industry.
See also
MIT Unveils Self-Distillation Fine-Tuning to Combat AI Catastrophic Forgetting
C&EN Reveals How AI and Funding Transform the Future of Chemistry Research
Kimi AI Launches Versatile Research Tool to Compete with ChatGPT and Claude
U of T and AMD Launch AI Research Lab with 100 Projects Over Three Years
LSU Experts Discuss $10B AI Data Centers’ Impact on Louisiana’s Energy Landscape
