Link Search Menu Expand Document
CS7670

CS7670: Neural Networks for Systems (Seminar)


10/27: lfs task details released

  • see section 4 of week8b’s note for task details and deliverables

Announcements

Course information

  • Course Number: CS 7670
  • Lectures: MonWed 14:50 - 16:30
  • Room: Hastings Suite 114
  • Instructor: Cheng Tan

Description

Deep learning (or neural networks) is popular and promising. It improves, and sometimes revolutionizes, many fields; for example, CV, NLP, and biology, to name a few. If you work in systems (like OSes, databases, and networked systems) and are now wondering, how about systems? This is the seminar for you.

This seminar studies neural networks for systems (we call it NN4Sys). NN4Sys is emerging and has potential. People have made many proposals for NN4Sys in the past five years. Some are revolutionary (my opinion), some are useful, and some are failed attempts.

In this seminar, we will study and discuss papers, some of which are classic systems papers (could be papers in the early days); others are recent papers using neural networks to solve the same or similar problem. We will read papers from research areas of operating systems, databases, and networked systems.

The goals of this seminar are the following:

  • We will study and explore NN4Sys, a young area in systems research. The course also aims to provide necessary tools for people who want to work on NN4Sys related problems.

  • All the participants of this seminar will work together on a collaborative final project to implement a system that leverages machine learning for better performance.

  • In addition, you might learn, as a bonus, how to read papers and think critically about papers.

Prerequisites and background

CS7670 is a seminar in systems. So,

  • we expect you to have basic systems background, for example, CS5600 or similar courses;
  • we expect you know how to program in Python;
  • we do not expect you to understand machine learning or neural networks (in fact, you will learn the basics of neural networks from this seminar).

The course

  • The lectures will cover topics of computer systems and neural networks for systems. See schedule here.
  • The final project will be a collaborative research project. See here.
  • The labs are warming up projects for you to get familiar with the background of the final project. See them here.
  • The announcements will be posted here.
  • The policies and grading for this course are here.