NVIDIA-NTU Artificial Intelligence and High-Performance Computing Application Research Program

Back

Title	NVIDIA-NTU Artificial Intelligence and High-Performance Computing Application Research Program
Office	NVIDIA-臺大人工智慧聯合研究中心
TIME	2024/11/20 上午 10:16:40
Content	1. Background: The establishment of the NVIDIA-National Taiwan University Artificial Intelligence Joint Research Center （NVIDIA-NTU AI Center） aims to promote the development and application of artificial intelligence research technology in Taiwan and foster talent. The primary goal of this call-for-project is to connect NVIDIAs high-performance GPU computing resources with National Taiwan Universitys interdisciplinary research teams, jointly pushing forward the development and application of artificial intelligence technology in Taiwan. It is hoped that this proposal can accelerate the development and realization of AI application technologies and simultaneously cultivate more outstanding talent in artificial intelligence. 2. Call for Topics: 2-1. Generative AI and Digital Twins Applications: Applications related to Generative AI and Digital Twins, including but not limited to its use in music, computer vision, finance, law, and other areas. 2-2. Large Language Models: Technical developments and research related to large language models in speech processing and natural language processing. Also, the application of large language model technologies in various practical scenarios. 2-3. Physics-Informed Neural Networks: Development of Physics-Informed Neural Networks （PINN） for scientific computation. This integrates the causal relationships described by governing partial differential equations （PDEs） in physics with simulation data from CAE solvers or observational data. Relevant in areas like fluid mechanics and electromagnetism. 2-4. Quantum Computing Simulation: Research in quantum computing simulation, for instance, assisting researchers in simulating quantum systems on classical computers to explore the potential of quantum computing. 2-5. Drug Discovery: Research related to drug development, such as leveraging deep learning for drug molecule design, protein structure prediction, and other aspects. 2-6. Healthcare: Medical-related research and applications, aimed at enhancing diagnostic and therapeutic outcomes. 2-7. Other Topics Related to NVIDIA SDK: Innovative applications developed using the technologies and tools provided by NVIDIA SDK, preferably those requiring large-scale computation. 3. Qualifications for the principal investigator: Assistant professors or higher from National Taiwan University. There are no qualification restrictions for co-principal investigators. 4. Application Method: From the announcement date until November 30, 2024, please email the application form （see attached file for details） to the following email address:　nvidia-ntu-applied-research@googlegroups.com 5. Review Process: The project application will be jointly reviewed by a committee composed of NVIDIA and National Taiwan University （NTU） members. The project review will be based on three main assessment criteria. 5-1. High-speed Computing Demand and Use of NVIDIA SDK: The review committee will assess the applications need for high-performance computing, prioritizing research applications that require extensive GPU computing resources, and how the application plan effectively utilizes the NVIDIA SDK in its research. 5-2. Project Impact: The review committee will evaluate the anticipated impact of the application plan on various fields such as science, engineering, or society. 5-3. Project Feasibility: This includes assessing whether the research plan is practically feasible, such as whether there is a feasible experimental design, sufficient time and resources to complete the research, etc. A total of five places are expected to be offered for this call-for-project. The committee will complete the review process and announce the results by January 31, 2025. The project can last up to 3 months, with a recommended start date no earlier than March 1, 2025. 6. Resource: Each selected research project will receive up to three months of computational resource assistance, according to the needs stated in its application and the review results. NVIDIAs professional Solution Architect （SA） team will offer technical guidance. During the execution of the entire core research plan, the SA team will provide technical support during business hours. To obtain the usage rights of computing resources, each selected research project must have at least one member become a contractor of NVIDIA. This member will act as the primary contact person between the team and NVIDIA and will be responsible for handling relevant usage rights matters. 7. Rights and Obligations 7-1. Each research project should cooperate as per the NVIDIA-NTU AI Centers requirements, assisting in understanding and monitoring the projects progress. This includes planning for academic research exchanges, scheduling, and updates on research progress. 7-2. Within two months of the projects completion, a final report must be submitted. This report should encompass detailed information about the entire project, including but not limited to its objectives, methods, implementation process, findings, and conclusions. The report has no specific format, but its length should be at most 20 pages. 7-3. Any research outcomes from the selected project, if published in international conferences, academic journals, newspapers, magazines, or exposed on social media, should include a note acknowledging the support from this program. If a collaboration with the NVIDIA SA team has occurred, they should be listed as co-authors in the publication. 7-4. For intellectual property rights arising from original innovative technologies of the NVIDIA-NTU AI Center or NVIDIA developed through this project, ownership issues will be determined based on the contributions of both parties and the pre-established agreement. 8. Matters needing attention This project provides one-time large-scale computing power, which is especially suited for training large models. However, this computing power type is unsuitable for the inference phase. When participating in this project, it is recommended that teams plan and prepare carefully. For example, teams can gather all the necessary data beforehand and run initial tests and adjustments on smaller models. This way, when they use the large-scale computing power, they can complete the training smoothly and make the most of this valuable resource.
Fee	Free
Active Date	2024/11/20 - 2024/11/30
Active Place	no
Related website	http://docs.google.com/document/d/1Rm5eIwMsDQ0ZTr_gm5oPzMXwS0DUuEC30vKgdQNEe-k/edit
Contact	NVIDIA-NTU Research Center
Tel	02-33663366#25531
EMail	ntunvidia@ntu.edu.tw
Reader	Faculty/Staff, Student, Alumni, Others

Back