ECE-GY 9483/CSCI-GA 3033

Special Topics in Electrical Engineering

EFFICIENT AI AND HARDWARE ACCELERATOR DESIGN

ECE-GY 9483 Spring 2025

The course will focus on recent advancements in the design of efficient neural networks, specifically on how to create and optimize AI models for improved performance, scalability, and resource efficiency. Students will explore key techniques like model compression, pruning, quantization, and model distillation for CNN, RNN, Transformer and LLM, aimed at reducing computational complexity and memory usage while maintaining accuracy. Additionally, the course will cover efficient training and inference methods, including distributed computing, parallelism, and low-precision computation, which are crucial for deploying AI on resource-limited devices such as smartphones or edge computing systems. Students will also study advanced hardware architectures of AI accelerators, gaining insights into the hardware implementation of neural network computations on these specialized systems.

 

Lecture Time: Wednesday 5:00-7:30pm EST

Lecture Location: Course Meeting Room

Office Hour: Wednesday 11:30am-12:30pm or by appointment

Office Location: Rm 1003, 370 Jay Street

Readings Required Text: Goodfellow, Ian. "Deep learning." (2016). https://www.deeplearningbook.org/

Evaluation Breakdown:

  • Participation (5%)

  • Assignments (30%): total three of them, each counts 10%

  • In-course presentation (10%) 

  • Midterm (25%) 

  • Final project (30%)

    • Proposal 5%

    • Final presentation 15%

    • Final report 10%

Contact:

  • For questions related to course, use efficientaiaccelerator@gmail.com

  • For other inquiries, personal matters, or emergencies, you can email me at sai.zhang@nyu.edu

  • If you are interested in getting updates, please sign up here to join our mailing list to get notified.

Instructor
Sai Zhang
Assistant Professor
Teaching Assistants
Zhenyuan Dong
Grader
Sai Zhang
Assistant Professor

Course Schedule

Date
Topic
Logistics

Jan 22

Lecture 1: Intro & Basic topics of DNN

 

Jan 29

Lecture 2: Convolutional Neural Networks & RNN

 

Feb 05

Lecture 3: Transformer and its Application in AIGC

 

Feb 12

Lecture 4: Pruning Strategies for Efficient DNN Implementation

Assignment 1 out

Feb 19

Lecture 5: DNN Quantization

 

Feb 26

Lecture 6: Distillation, Low Rank Decomposition and Neural Architecture Search 

Assignment 1 due

Assignment 2 out

Mar 05

Lecture 7: Efficient DNN Training

 

Mar 12

Lecture 8: Machine Learning System for Training and Inference

Assignment 2 due

Assignment 3 out

Mar 19

Lecture 9: Intro to AI Accelerator

Project proposal due

In-class midterm exam

Mar 26

Spring Break

 

Apr 02

Lecture 10: CNN Dataflow & Hardware Accelerators

Assignment 3 due

Apr 09

Lecture 11: Transformer & LLM Accelerators

 

Apr 16

Lecture 12: Hardware Accelerator for DNN Training

 

Apr 23

Lecture 13: New Computation Paradigms

Invited Talk

Apr 30

Lecture 14: New playground for Efficient AI: AR/VR

Invited Talk

May 14

Final Presentation

 

Collaborators