𝐘𝐚𝐥𝐞 𝐭𝐨 𝐁𝐮𝐢𝐥𝐝 𝐚 𝐇𝐲𝐛𝐫𝐢𝐝 𝐐𝐮𝐚𝐧𝐭𝐮𝐦-𝐂𝐥𝐚𝐬𝐬𝐢𝐜𝐚𝐥 𝐆𝐏𝐓 𝐟𝐨𝐫 𝐍𝐨𝐯𝐞𝐥 𝐌𝐨𝐥𝐞𝐜𝐮𝐥𝐞 𝐆𝐞𝐧𝐞𝐫𝐚𝐭𝐢𝐨𝐧 𝐔𝐬𝐢𝐧𝐠 𝐍𝐕𝐈𝐃𝐈𝐀 𝐂𝐔𝐃𝐀-𝐐

Wednesday, November 20, 2024

Yale researchers, in collaboration with NVIDIA, are pioneering a hybrid quantum-classical approach to generative machine learning for novel molecule generation. Leveraging the transformer architecture’s self-attention mechanism, the team aims to accelerate drug discovery by efficiently exploring chemical spaces using quantum computing. Their innovative algorithm integrates transformer-based models with quantum devices, overcoming limitations of current quantum machine learning frameworks. While quantum hardware is not yet ready for real-world implementation, simulations using NVIDIA’s CUDA-Q platform demonstrate the potential of encoding sequence elements as quantum states. This groundbreaking work, which bridges quantum computing and generative AI, is expected to be announced at The International Conference for High Performance Computing, Networking, Storage, and Analysis (SC 2024) and released on the arXiv preprint server in the coming months.

Further information at this link.