集思EE电子工程AI芯片专题

来源：北京集思未来科研辅导时间：2023/8/10 10:46:16

集思EE电子工程AI芯片专题：电子工程/电子与计算机工程/集成电路/人工智能/微电子学

密集项目：EE电子工程 AI芯片专题：人工智能ChatGPT核心硬件基石研究 基于FPGA、GPU的大算力集成电路设计【大学组】

开始日期: 2023-08-06

课时安排: 4周在线小组科研学习+2周不论文指导学习。

Prerequisites

适合人群

适合年级 (Grade): 大学生及以上

适合专业 (Major): 电子工程、电子与计算机工程、自动化等专业或者希望修读相关专业的学生；学生需要具备数学、信号处理、VLSI 和计算机体系结构等相关基础

Instructor Introduction

导师介绍

Dejan

加州大学洛杉矶分校 (UCLA)

终身正教授

Dejan导师现任加州大学洛杉矶分校（UCLA）电子计算机工程终身正教授、电气与电子工程师协会会士(IEEE Fellow, Class of 2021）、美国医学与生物工程院院士(AIMBE Fellow， Class of 2022)，获得美国自然科学基金（NSF）早期职业奖（Early Career Development Awards)、ISSCI杰出论文奖，拥有加州大学伯克利分校博士学位。Dejan导师的研究聚焦植入式神经调节系统、领域特定计算、设计方法论，在国际期刊发表论文110余篇，h指数48，i10指数135，引用量高达8500+。

Prof. Dejan (IEEE Fellow、AIMBE Fellow) received the Ph.D. degree from the University of California at Berkeley in 2006., He is currently a Professor of electrical and computer engineering at the University of California at Los Angeles (UCLA). He is also affiliated with UCLA Bioengineering Department, Neuroengineering field. He is also a Co-Director of the Interdisciplinary Center for Neurotechnology (CENT) Prof.Dejan received the NSF CAREER Award in 2009 and the 2014 ISSCC Lewis Winner Award for Outstanding Paper. He also received the 2007 David J. Sakrison Memorial Prize for his Ph.D. degree.

任职学校

加州大学洛杉矶分校（UCLA）始建于1919年，是一所拥有百年历史的世界公立研究型大学，培养了14名诺贝尔奖获得者，是美国“新常春藤”名校之一。被《美国新闻与世界报道》《泰晤士高等教育》《华尔街日报》等多家放心报刊评为美国公立大学 ，被评为2018《福布斯》较具价值大学排名全美第1位，2020QS毕业生就业力排名世界第3位 。2021-22年度，UCLA位列软科世界大学学术排名第13位，U.S. News世界大学排名第14位，泰晤士高等教育世界大学排名第20位。

Program Background

项目背景

AI的关键基础要素是数据、算法和算力。随着云计算的广泛应用，特别是深度学习成为当前AI研究和运用的主流方式，AI对于算力的要求不断进步。具有海量并行计算能力、能够加速AI计算的AI芯片应运而生。随着技术成熟化，Al芯片的应用场景除了在云端及大数据中心，也会随着算力逐渐向边缘端移动，部署于智能家居、智能制造、智慧金融等领域；同时还将随着智能产品种类日渐丰富，部署于智能手机、安防摄像头、及自动驾驶汽车等智能终端，智能产品种类也日趋丰富。以GPU、FPGA、ASIC为代表的AI芯片，是目前可大规模商用的技术路线，是AI芯片的主战场。

The three key fundamental elements of AI are data, algorithms and computing power. With the widespread use of cloud computing, especially deep learning has become the mainstream way of AI research and application, the requirements of AI for computing power keep rising rapidly. AI chips with massive parallel computing capability and the ability to accelerate AI computing have emerged. As the technology matures, the application scenario of Al chips will not only be in the cloud and big data center, but will also be deployed in smart home, smart manufacturing, smart finance and other fields as the computing power gradually moves to the edge; meanwhile, it will also be deployed in smart terminals such as smart phones, security cameras, and self-driving cars as the variety of smart products becomes richer. AI chips represented by GPU, FPGA and ASIC are the main battlefield of AI chips as they are the technology route that can be commercially used on a large scale at present.

Program Description

项目介绍

本项目将研究如何使用现有硬件（如FPGA、GPU、CPU等）加速人工智能AI或机器学习ML的计算，并了解新的AI/ML加速架构，涉及算法、信号处理、超大规模集成电路和计算机体系结构等多个领域的内容。通过本项目的学习，能够帮助学生了解AI/ML算法的硬件成本，以及如何在各种硬件目标上映射AI/ML算法，并能够优化相关硬件设计，更大程度地支撑对AI/ML的加速作用。项目结束时提交项目报告，进行成果展示。This project will investigate how to accelerate AI AI or machine learning ML computations using existing hardware (e.g. FPGAs, GPUs, CPUs, etc.) and understand new AI/ML acceleration architectures in a variety of areas including algorithms, signal processing, exascale integrated circuits, and computer architecture. This project will help students to understand the hardware cost of AI/ML algorithms and how to map AI/ML algorithms on various hardware targets and be able to optimize the related hardware design to support the acceleration effect on AI/ML to a greater extent. A project report will be submitted at the end of the project to present the results.

Syllabus

项目大纲

计算机体系结构与数字算术（CPU, GPU, FPGA, ASIC） Compute architectures, and digital arithmetic (number representations)

基于FPGA、CPU/GPU+的硬件加速器及SoC框架 Understand SoC approach with FPGA, CPU/GPU + hardware accelerators

深度学习DL与图神经网络GNN算法基础 Deep learning (DL), graph neural networks (GNNs)

RTL和描述：基于Python库的描述实现硬件加速 RTL and high-level descriptions

能够在域内重新配置运行的新型计算机（AI/ML）New class of computers capable of runtime reconfiguration within a domain (AI/ML)

项目回顾与成果展示 Program Review and Presentation

论文辅导 Project Deliverables Tutoring

Program Outcome

项目收获

4周在线小组科研学习+2周不论文指导学习 共125课时

项目报告

学员获主导师Reference Letter

EI/CPCI/Scopus/ProQuest/Crossref/EBSCO或同等级别索引国际会议全文投递与发表指导（可用于申请）

结业证书

成绩单

尊重原创文章，转载请注明出处与链接：http://www.peixun360.com/6861/news/649929/违者必究！ 以上就是北京集思未来科研辅导 小编为您整理 集思EE电子工程AI芯片专题的全部内容。