Towards Best Possible Deep Learning Acceleration on the Edge – A Compression-Compilation Co-Design Framework
The lecture presented by Assistant Professor Yanzhi Wang introduces a compression‑compilation co‑design framework (CoCoPIE) that achieves real‑time deep‑learning inference on edge devices through novel pruning and quantization techniques, delivering up to 180× speedup without accuracy loss.
Event: AI Scientific Frontiers Lecture Series (第7讲), February 7, 2021, 10:00–11:00 (online via Tencent Meeting).
Speaker: Yanzhi Wang, Assistant Professor, Dept. of ECE, Northeastern University, Boston, MA. He received his B.S. from Tsinghua University (2009) and Ph.D. from USC (2014). His research focuses on model compression and platform‑specific acceleration of deep learning applications, achieving state‑of‑the‑art compression rates and energy‑efficient hardware implementations.
Abstract: Mobile and embedded devices are key carriers of deep learning, yet real‑time DNN inference on edge is challenged by limited compute and storage. Existing compression methods trade off accuracy and hardware performance.
CoCoPIE (Compression‑Compilation Co‑Design) introduces fine‑grained structured pruning (pattern‑based, block‑based) and a novel quantization scheme that attain high hardware performance comparable to aggressive pruning while preserving zero accuracy loss, thanks to compiler support. The framework enables real‑time execution of tasks such as object detection, pose estimation, activity detection, and speech recognition on off‑the‑shelf mobile devices, achieving up to 180× speedup over prior work.
Demonstrations are available at: https://space.bilibili.com/573588276?from=search&seid=3414392881179119028
Signed-in readers can open the original source through BestHub's protected redirect.
This article has been distilled and summarized from source material, then republished for learning and reference. If you believe it infringes your rights, please contactand we will review it promptly.
DataFunTalk
Dedicated to sharing and discussing big data and AI technology applications, aiming to empower a million data scientists. Regularly hosts live tech talks and curates articles on big data, recommendation/search algorithms, advertising algorithms, NLP, intelligent risk control, autonomous driving, and machine learning/deep learning.
How this landed with the community
Was this worth your time?
0 Comments
Thoughtful readers leave field notes, pushback, and hard-won operational detail here.