5/22 (Thursday) Deep Learning, Deep Research, Continuous Learning, and NPU Processors @ 陽明交大工程四館 EDB27
Title: Deep Learning, Deep Research, Continuous Learning, and NPU Processors
Speaker: Stephen Huang, Founder of TranXform AI
Time/Location: 5/22 (Thursday), 10am, 工程四館 EDB27
Register: https://forms.gle/
Abstract:
This talk delves into the interconnected domains of deep learning, deep research, continuous learning, and NPU processors, targeting college, master's, and PhD students. We'll begin with foundational deep learning principles, transition to the methodologies of rigorous research, and emphasize strategies for continuous learning in this rapidly evolving field. Finally, we will explore the architecture and impact of Neural Processing Unit (NPU) processors on accelerating deep learning workloads. This session aims to provide a comprehensive understanding of deep learning's journey, from theoretical underpinnings to practical research and specialized hardware implementation, equipping attendees with insights relevant to all levels of expertise.
Short introduction to the speaker:
For the past 32 years, I've been deeply involved in the evolution of computing hardware and software, with a focus on accelerating artificial intelligence. My experience spans GPU drivers, algorithms, and hardware architecture at S3 Graphics. I've worked on CPU and interconnect design at Intel, NPU compiler architecture at Apple, and NPU architecture at Amazon. More recently, I led computer vision research and NPU architecture at Oppo, and I'm currently the CEO and Founder of Tranxform.
About Tranxform
AI: At Tranxform AI, we're dedicated to developing high-level computing processors designed to accelerate a wide range of AI models, including CNNs, RNNs, LSTMs, Transformers, Diffusion models, and Diffusion LLMs. Our goal is to deliver an end-to-end solution, encompassing algorithm, compiler, driver, hardware design, and application development, providing a turnkey solution for applications in robotics, autonomous vehicles, drones, medical technology, and healthcare.
