CSE 290D: Seminar on Integrative AI Engineering

Seminar Overview and Goals

Machine learning, data science, and more generally, software and systems powered by artificial intelligence (AI) techniques have led to a new era of computing applications that augment human capabilities, automate many mundane tasks, and permeate the virtual and physical worlds. These applications, which have captured the public imagination, range from benign ones such as recommendation systems, personal conversational assistants, healthcare assistants and robotics, Internet of Things and predictive maintenance, and self-tuning software systems to controversial ones such as autonomous vehicles, AI-enabled mass surveillance, and AI-powered weapons of war.

Such applications require holistic thinking blending computer science and engineering with math, physics, cognitive science, ethics, and various engineering fields, as well as the application domain, leading to an emerging AI-focused engineering discipline that is often called "Integrative AI". In a sense, ML/AI plays a similar role in this emerging engineering field that physics does to electrical, mechanical, and many other existing engineering fields. This field is transforming the way researchers and practitioners work in the sciences and engineering and the way companies obtain business insights, creating entirely new market categories, and even changing the way society itself functions.

The goal of this new seminar is to explore and understand this dramatic emerging space of integrative AI engineering through the lens of high-profile applications and the science and engineering behind them. Students will also acquire the skills of reading papers in this area and presenting them to a technical audience.

Administrivia

Seminar Meetings: Mon 4:00-4:50pm; CSE/EBU3b 4140

Organizer: Arun Kumar; Office: CSE 3218

Seminar Content and Format

  • There is a weekly reading list of articles and papers on a variety of topics under the theme of Integrative AI Engineering. Students are expected to read all papers in the reading list. Each student has to present one paper from the list in class and steer a discussion about the paper. Students will be randomly assigned to papers by the organizer.

  • Each seminar will have 3 presentations under one topic. There will be 9 seminars in total this quarter.

  • If more than 27 students are enrolled, some presentations will be by pairs of students (randomly assigned by the organizer).

  • Students are free to chose how they want to present: slidedeck or whiteboard. The content should be well-organized, readable, and well-presented.

Pre-requisites

Basic courses on machine learning, data mining, databases, and systems will be helpful but are not really required. Students should also be willing to learn the basics of new application domains on their own.

Grading

  • This is a 1-credit pass-fail seminar.

  • The requirement for a pass grade is to present one of the papers/articles (randomly assigned by the organizer) and also attend at least 6 of the other meetings. The organizer will be marking attendance in class.

Classroom Rules

  • You are encouraged to ask questions and participate in in-class discussions. Please raise your hand before asking questions during a talk.

  • Harassment or intimidation of any form against any student will not be tolerated in class.