Colloquium Series

Once every week while school is in session, EECS invites a distinguished researcher or practitioner in a computer science or electrical and computer engineering-related field to present their ideas and/or work. Talks are generally targeted to Electrical Engineering and Computer Science graduate students. This colloquium series is free and open to everyone.

Upcoming Colloquia

ASU Connection One Center on Analog/RF and Power Management IC

Monday, April 28, 2014 - 4:00pm - 4:50pm
KEC 1001
Sayfe Kiaei
School of Electrical, Computer, and Energy Engineering
Arizona State University

The first part of this presentation will give a brief overview of research at Connection One on RF, Analog and PMIC. This will be followed by a presentation on Digital Linear Drop-out Regulators and the development of Power Management IC. The development of multi-core highly integrated systems-on-a-chip has created the need for small, fully integrated voltage regulators that operate on a per-core basis. In order to maximize efficiency, most SOC's apply dynamic voltage and frequency scaling (DVFS) on each block of the system to adjust the power based on performance demands. Analog regulators are poorly suited to this task as they are difficult to integrate on sub-micron processes, consume more power, and require precision external capacitors to ensure stability. The development of Digital LDO regulators is intended to address these drawbacks of analog regulators.

Speaker Biography: 

Dr. Sayfe Kiaei is has been with ASU since January 2001. He is a Professor and the Director of the Connection One Center (NSF I/UCRC Center), and Motorola Chair in Analog and RF Integrated Circuits. From 1993 to 2001, he was a Senior Member of Technical Staff with the Wireless Technology Center and Broadband Operations at Motorola where he was responsible for the development of RF & Transceiver Integrated Circuits, GPS RF IC, and Digital Subscriber Lines (DSL) transceivers. Before joining Motorola, Dr. Kiaei was a Professor at Oregon State University from 1987-1993 where he taught courses and performed research in digital communications, VLSI system design, advanced CMOS IC design, and wireless systems. Dr. has published more than 200 journal and conference papers and holds several patents. Dr. Kiaei is an IEEE Fellow and a member of IEEE Circuits and Systems Society, IEEE Solid State Circuits Society, and IEEE Communication Society. Dr. Kiaei has been organizer, on the technical program committee and/or chair of many conferences, including: RFIC, MTT, ISCAS, and other international conferences.

Machine Teaching: Frenemy of Machine Learning

Monday, May 12, 2014 - 4:00pm - 4:50pm
KEC 1001
Xiaojin (Jerry) Zhu
Associate Professor
Department of Computer Sciences
University of Wisconsin-Madison

Consider the inverse problem of machine learning: a teacher knows a learner's learning algorithm and wants to construct the smallest (non-iid) training set to guide the learner to a specific target model.  This problem, which we call machine teaching, is about designing the optimal "lesson" to maximally influence the learner.  One application of machine teaching is in the security of machine learning systems that accept online training data.  Here, the teacher is an attacker who can manipulate the training data. The computational problem is for the attacker to identify the minimum-cost manipulation so that the machine learner will be misled to a model that is beneficial to the attacker.  A more friendly application of machine teaching is in education, where the teacher wishes to design the best lesson for a human student.  Note that in both applications the teacher may only interact with the learner via the training data.  I will introduce an optimization-based framework for machine teaching, balancing the goals of "teaching well" and "minimizing teaching effort."  For certain learners, machine teaching has a closed-form solution.  But in general the optimization problem is combinatorial.  I will discuss two approximate solution techniques based on conjugate duality and submodularity. I will also discuss the relation between machine teaching, active learning, and teaching dimensions. Finally, I demonstrate the application of machine teaching with attacks on several popular machine learning models.

Speaker Biography: 

Xiaojin Zhu is an Associate Professor in the Department of Computer Sciences at the University of Wisconsin-Madison.  Dr. Zhu received his B.S. and M.S. degrees in Computer Science from Shanghai Jiao Tong University in 1993 and 1996, respectively, and a Ph.D. degree in Language Technologies from Carnegie Mellon University in 2005.  He was a research staff member at IBM China Research Laboratory from 1996 to 1998.  Dr. Zhu received the National Science Foundation CAREER Award in 2010, and best paper awards at ICML, ECML/PKDD, and SIGSOFT.  His research interest is in machine learning, with applications in natural language processing, cognitive science, and social media analysis.

Past Colloquia

Monday, April 21, 2014 - 4:00pm - 4:50pm
Jim Kemerling and Darrell Teegarden
Triad Semiconductor, Founder, CTO, Executive VP (Jim Kemerling)
Mentor Graphics, Business Unit Director, System Modeling & Analysis (Darrell Teegarden)
Friday, April 18, 2014 - 12:00pm - 1:00pm
Douglas W. Fisher
Vice President
General Manager, Software and Services Group
Intel Corporation
Friday, April 18, 2014 - 8:45am - 10:00am
Gregor Richards
PhD candidate
Secure Software Systems Lab
Purdue University
Friday, April 11, 2014 - 8:45am - 10:00am
Abhishek Jain
Postdoctoral Researcher
Cryptography and Information Security Group
Thursday, April 10, 2014 - 8:45am - 10:00am
Lizhong Chen
University of Southern California
Wednesday, April 9, 2014 - 8:45am - 10:00am
Jay McCarthy
Assistant Professor
Computer Science Department
Brigham Young University
Thursday, April 3, 2014 - 8:45am - 10:00am
Jinsub Kim
Postdoctoral Associate
School of Electrical and Computer Engineering
Cornell University