Lazy Aggregation for Communication-Efficient Collaborative Machine Learning 🗓 🗺

register>

Date: June 24 2019.
Time: 04:30 PM to 06:00 PM (EDT)
Speaker: Mark Barlow

Location:

6600 Washington Ave. S.
Eden Prairie, Minnesota
United States 55344

Cost: Free
RSVP: Required.
Event Details & Registration: (URL)

Summary:

Considering the massive number of devices, centralized machine learning via cloud computing incurs considerable overhead, and raises serious privacy concerns. Today, the consensus is that future machine learning tasks have to be performed starting from the network edge, namely devices. In this context, we will highlight key challenges in learning at the edge, including communication overhead, heterogeneity, and adversarial attacks. Wedding optimization techniques with system-level considerations, we will introduce novel methods for solving distributed learning problems. Our methods are simple to implement, and come with rigorous performance guarantees.

Biography:

Tianyi Chen received the B. Eng. degree in Communication Science and Engineering from Fudan University, the M.Sc. and Ph.D degrees in Electrical and Computer Engineering (ECE) from the University of Minnesota (UMN), in 2014, 2016 and 2019, respectively. Starting in August 2019, he will with Department of ECSE at Rensselaer Polytechnic Institute (RPI) as an Assistant Professor. Between 2017 and 2018, he was a visiting scholar at Harvard University, University of California, Los Angeles, and University of Illinois Urbana-Champaign.
His research interests lie in optimization and machine learning with applications to large-scale networked systems such as Internet-of-Things, next-generation computing systems, and energy systems. He was a Best Student Paper Award finalist in the 2017 Asilomar Conf. on Signals, Systems, and Computers. He received the National Scholarship from China in 2013, the UMN ECE Department Fellowship in 2014, and the UMN Doctoral Dissertation Fellowship in 2017.

Models and Inference with Network Structure 🗓 🗺

Date: Thursday, June 20, 2019
Time: 11:00 AM – 1:00 PM (EDT)
Speaker: Brandon Oselio
Location: EECS 3316
Cost: none
RSVP: none
Event Details & Registration: http://www.eecs.umich.edu/eecs/etc/events/showevent.cgi?5165

Abstract: Complex, structured data is ubiquitous in both industrial and academic settings and has elicited a commensurate interest in utilizing structured data to inform inference and decisions. We are particularly interested in data that has network structure and on problems that benefit from the application of network-based algorithms. We focus on four research problems of interest: scalable and realistic models for network valued data, graph-based estimation of information theoretic quantities, summarization of complex time-varying data using dynamic graphs, and finally community detection on large multi-layer networks. This work advances the state-of-the-art in several directions. First, it introduces a new framework for complex hierarchical network interaction data using the concept of edge exchangeability. Second, it obtains new tight bounds for the multi-class Bayes error rate based on a graph-based technique, specifically the minimal spanning tree. Third, it introduces a new estimation method for Henze-Penrose divergence, a quantity relevant for graph-based multi-class classification. Fourth, it introduces adaptive directed information for estimating directed interaction networks. Fifth, it provides a comprehensive approach to multi-layer network community detection. Throughout, examples are provided using real datasets, such as the Enron email dataset, an arXiv dataset, and Twitter.

Exploring the Life and Achievements of Nikola Tesla 🗓 🗺

register>

Date: June 28 2019.
Time: 06:00 PM to 08:30 PM (EDT)
Speaker: Mark Barlow

Location:
214 N. Trenton Ave.
Pittsburgh, Pennsylvania
United States 15221

Cost: Free
RSVP: Required.
Event Details & Registration: (URL)

Summary:

Join us in appreciating the life and achievements of electrical engineer Nikola Tesla! We will reflect on Tesla’s past, recognize his contributions to the present, and consider his warnings about the future. The presentation will touch on aspects of Tesla’s life, the impact of AC power, and the Tesla Coil.
And YES…… This presentation will include “Live” Tesla Coil demonstrations!
A tour of the Protohaven facility will also be available a few minutes before the presentation starts!

Biography:

Mr. Mark Barlow has 8 years experience in the energy industry working with power electronics, solar photovoltaic grid tied and battery storage power systems. His years of experience include employment with solar equipment integrators, applications engineering, and utility analytics.

Mark established his own company DC to Power, LLC in 2011 to provide technical assistance on energy projects with the ambition of making solar energy more cost effective.

Additionally, Mark has 3 years of reliability engineering experience from the semiconductor industry; he has been a member of the IEEE for 16 years and a Tesla Coil Builder for the last 22 years.

Mr. Barlow received his Masters of Science in Engineering from Youngstown State University where he completed his thesis on the topic of fabricating Schottky Diodes on Silicon Carbide.

Taking Bipedal Robots from Science Fiction to Science Fact: Trials and Tribulations 🗓 🗺

register>

Date: Jun 05 2019.
Time: 06:00 PM to 08:00 PM (EDT)
Speaker: Prof. Jessy Grizzle of University of Michigan.

Location:
2350 Hayward St
Ann Arbor, Michigan
United States 48109

Cost: Free
RSVP: Required.
Event Details & Registration: (URL)

Summary:

The fields of control and robotics are working hand-in-hand to development bipedal machines that can realize walking motions with the stability and agility of a human. This talk will show how model-based feedback control and optimization methods are enhancing the ability to achieve highly dynamic locomotion in bipedal robots and exoskeletons. The presentation will be amply illustrated with graphics and videos of our experiments to make the material accessible to a wide audience. Just for fun, a few equations will be thrown in!

Biography:

Jessy Grizzle

Jessy W. Grizzle received the Ph.D. in electrical engineering from The University of Texas at Austin in 1983 and in 1984 held an NSF-NATO Postdoctoral Fellowship in Science in Paris, France. Since September 1987, he has been with The University of Michigan, Ann Arbor, where he is the Elmer G Gilbert Distinguished University Professor and the Jerry and Carol Levin Professor of Engineering. He jointly holds sixteen patents dealing with emissions reduction in passenger vehicles through improved control system design. Professor Grizzle is a Fellow of the IEEE and of IFAC. He received the Paper of the Year Award from the IEEE Vehicular Technology Society in 1993, the George S. Axelby Award in 2002, the Control Systems Technology Award in 2003, the Bode Lecture Prize in 2012, and the IEEE Transactions on Control Systems Technology (TCST) Outstanding Paper Award in 2014. His work on bipedal locomotion has been the object of numerous plenary lectures and has been featured in The Economist, Wired Magazine, Discover Magazine, Scientific American, Popular Mechanics and several television programs. Since September 2016, he has been the Director of Michigan Robotics.  You can follow the construction of our beautiful new building

Neural Interfaces for Controlling Finger Movements 🗓 🗺

register>

Date: June 25 2019.
Time: 06:00 PM to 08:00 PM (EDT)
Speaker: Prof. Cynthia Chestek of  University of Michigan.

Location:

2350 Hayward St
Ann Arbor, Michigan
United States 48109

Cost: Free
RSVP: Required.
Event Details & Registration: (URL)

Summary:

Brain machine interfaces or neural prosthetics have the potential to restore movement to people with paralysis or amputation, bridging gaps in the nervous system with an artificial device. Microelectrode arrays can record from hundreds of individual neurons in motor cortex, and machine learning can be used to generate useful control signals from this neural activity. Performance can already surpass the current state of the art in assistive technology in terms of controlling the endpoint of computer cursors or prosthetic hands. The natural next step in this progression is to control more complex movements at the level of individual fingers. Our lab has approached this problem in three different ways. For people with upper limb amputation, we acquire signals from individual peripheral nerve branches using small muscle grafts to amplify the signal. After a successful study in animals, human study participants have recently been able to control individual fingers online using indwelling EMG electrodes within these grafts. For spinal cord injury, where no peripheral signals are available, we implant Utah arrays into finger areas of motor cortex, and have successfully decoded flexion and extension in multiple fingers. Decoding “spiking band” activity at much lower sampling rates, we recently showed that power consumption of an implantable device could be reduced by an order of magnitude compared to existing broadband approaches, and fit within the specification of existing systems for upper limb functional electrical stimulation. Finally, finger control is ultimately limited by the number of independent electrodes that can be placed within cortex or the nerves, and this is in turn limited by the extent of glial scarring surrounding an electrode. Therefore, we developed an electrode array based on 8 um carbon fibers, no bigger than the neurons themselves to enable chronic recording of single units with minimal scarring. The long-term goal of this work is to make neural interfaces for the restoration of hand movement a clinical reality for everyone who has lost the use of their hands.

Biography:

Cynthia Chestek

Cynthia A. Chestek received the B.S. and M.S. degrees in electrical engineering from Case Western Reserve University in 2005 and the Ph.D. degree in electrical engineering from Stanford University in 2010. From 2010 to 2012, she was a Research Associate at the Stanford Department of Neurosurgery with the Braingate 2 clinical trial. She is now an associate professor of Biomedical Engineering at the University of Michigan, Ann Arbor, MI, where she joined the faculty in 2012. She runs the Cortical Neural Prosthetics Lab, which focuses on brain and nerve control of finger movements as well as to high-density carbon fiber electrode arrays. She is the author of 34 full-length scientific articles. Her research interests include high-density interfaces to the nervous system for the control of multiple degree of freedom hand and finger movements.

Nanoscale Devices based on Two-dimensional Materials and Ferroelectric Materials 🗓 🗺

Date: Thursday, May 30, 2019
Time: 2:00 PM – 3:00 PM (EDT)
Speaker: Dr. Wenjuan Zhu, Assistant Professor, University of Illinois

Location: 1003 EECS
Cost: none
RSVP: none
Event Details & Registration: http://www.eecs.umich.edu/eecs/etc/events/showevent.cgi?5152

Summary: Further scaling of complementary metal-oxide-semiconductor (CMOS) dimensions will soon lead to a tremendous rise in power consumption while limited gain in the performance of integrated circuits. “Beyond-CMOS” devices, based on new materials, device concepts and architectures, can potentially overcome these limitations and further improve the performance, reduce energy consumption, and add novel functionalities to the CMOS platform. In this talk, I will present nanoscale electronic and photonic devices based on two-dimensional (2D) materials and ferroelectric materials. In particular, I will discuss the logic devices, RF devices, photodetectors, plasmonic devices, and tunneling devices based on graphene and transition metal dichalcogenides. I will also present our recent results on non-volatile memories and ferroelectric tunneling junctions (FTJs) based on ferroelectric hafnium oxide and 2D ferroelectric indium selenides.


Wenjuan Zhu is an assistant professor in Department of Electrical and Computer Engineering at University of Illinois at Urbana-Champaign. Her current research interests are nanoscale electronic and optoelectronic devices based on two-dimensional materials (including transition metal dichalcogenides, graphene, and black phosphorus), and ferroelectric materials. She is a recipient of IBM faculty Award (2018), National Science Foundation CAREER Award (2017), Outstanding Technical Achievement Award in IBM (2008), Henry Prentiss Becton Graduate Prize for Exceptional Achievement in Research in Engineering and Applied Science at Yale University (2003). 
Prof. Zhu received her Ph.D. degree in the Department of Electrical Engineering at Yale University in 2003. After graduation, she joined IBM as an advisory Engineering/Scientist at Semiconductor Research and Development Center (SRDC) (2003-2008) and later as a Research Staff Member at T. J. Watson Research Center (2008-2014). In 2014, she joined the faculty at the University of Illinois and established a research group focusing on two-dimensional (2D) materials and nanoscale devices. Her research in the past has resulted in more than 100 publications in journals/conferences and more than twenty patents.

Energy-Efficient Mobile Computer Vision and Machine Learning Processors 🗓 🗺

Date: Wednesday, May 29, 2019
Time: 9:00 AM – 11:00 AM (EDT)
Speaker: Ziyun Li
Location: EECS 1005
Cost: none
RSVP: none
Event Details & Registration: http://www.eecs.umich.edu/eecs/etc/events/showevent.cgi?5151

Abstract: Technology scaling has driven computing devices to be faster, cheaper, and smaller while consuming less power in past decades. However, as technology scaling has become increasingly difficult in recent years, power has become the major constraint in performance, and thus, the improvement in the performance of mobile devices has begun to diminish. Moreover, emerging intelligent mobile systems are demanding increasing computing power. In light of this challenge associated with artificial intelligence, domain-specific architectures are widely believed to be the path to realizing considerable improvements in the efficiency, performance and cost of intelligent mobile systems. This dissertation presents several algorithm, architecture and circuit co-optimized solutions for intelligent and autonomous mobile systems, including vision-based stereo depth, optical flow, simultaneous localization and mapping (SLAM) and convolutional neural network- (CNN)-based image recognition. Together, these solutions enable the mobile systems to form a geometric and semantic understanding of the environment. Various optimizations including parallelism, scheduling, exploiting sparsity and circuit customization are applied to overcome the complexity of these problems for energy-efficient, real-time, robust operation.

Modeling of Downstream and Direct Plasma Systems for Highly Selective and Anisotropic Etching 🗓 🗺

Date: Tuesday, May 28, 2019
Time: 2:00PM – 4:00PM (EDT)
Speaker: Shuo Huang
Location: EECS 2311
Cost: none
RSVP: none
Event Details & Registration: http://www.eecs.umich.edu/eecs/etc/events/showevent.cgi?5138

Summary: The pursuit of higher integration has brought the semiconductor industry into the realm of nanoelectronics (e.g., 14 nm FinFET) and 3-dimensional structures (e.g., vertical NAND). Increasing challenges on selectivity and anisotropy have imposed stringent requirements on controlling low temperature plasmas for material processing. In this thesis, integrated reactor and feature scale modeling was performed for optimizing plasma etching process, with updates implemented into the Hybrid Plasma Equipment Model (HPEM) to investigate plasma properties and the Monte Carlo Feature Profile Model (MCFPM) to predict etch profiles. Highly selective etching of Si3N4 was achieved using downstream etch system consisting of a remote plasma source sustained in Ar/NF3/O2 mixtures, a plenum and a downstream chamber. Plasma is mainly confined to source region with a weak ion-ion plasma sustained afterglow. Dominant F and NO radicals (etchants of Si3N4) flow downstream and iteratively remove Si and N surface subsites. Highly anisotropic etching of high aspect ratio (AR) features in SiO2 with AR up to 80 was achieved using multi-frequency capacitively coupled plasmas sustained in Ar/C4F8/O2 mixtures. As AR increases, the dominant etching mechanism transitions from chemical sputtering to physical sputtering as the fluxes of energetic species (ions and hot neutrals) to etch front surpass those of conduction constrained polymerizing CFx and CxFy radicals.

The History of the 6502 Processor.. Or How Your Spreadsheet Died From Dysentery… 🗓 🗺

register>

Date: July 17 2019.
Time: 05:00 PM to 06:30 PM (EDT)
Speaker: Ben Leadholm

Location:
3537 Zenith Avenue South
Minneapolis, Minnesota
United States 55416

Cost: Free
RSVP: Required.
Event Details & Registration: (URL)

Summary:

Some of us are old enough to remember our first computer being the one of the Atari Game Systems, Atari 400 or 800, some form of Apple II, or the Commodore Vic 20 or C64.

There are fascinating histories behind the creation of the heart of most personal processing systems back in the late 70s and early to mid 80s.

We won’t focus on Intel’s 8080, the heart of the Altair, the first personal computer, and the computer that helped to spawn the HomeBrew Computer Club. Instead, we’ll start with Chuck Peddle, a microchip designer who left Motorola because the company didn’t want his idea of a stripped-down processor to eat market share away at their $300 68XX chip family.

Chuck jumps to MOS Technologies, and develops the 6502 Microprocessor, on sale for $25 in quantities of one.

It is this price point that attracts Steve Wozniak to build a working personal computer on plywood to plug into a TV called the Apple I, to show off at the HomeBrew Computer Club. Fifty Apple I circuit boards were sold for $666.66 each (roughly $3k today). Woz’s friend Steve Jobs thinks they could make a business selling pre-assembled computers, “1000 a week.”

There will be tales of intrigue, backstabbing, dirty tactics, and cutthroat competition from Steve Jobs, Bill Gates, Nolan Bushnell, Jack Tramiel, and Bil Herd.

There may be 8-bit machines on display (Atari | Commodore | Apple).

Some time may be allocated to let others reminisce about their ‘6502 moments.

Biography:

A developer for almost 25 years, starting out with Excel macros and parlaying those skills to an entry-level development position. From VB3 to Ruby (with a stint of .NET and Java in between). Ben is now contracting with specializations in Ruby, Python, and database migration.

A Distributed Energy Management Strategy For Resilient Shipboard Power System 🗓 🗺

register>

Date: May 30 2019.
Time: 05:30 PM to 07:00 PM (EDT)
Speaker: Kexing Lai

Location:
8100 Walton Parkway
New Albany, Ohio
United States 43054

Cost: Free
RSVP: Required.
Event Details & Registration: (URL)

Summary:

The shipboard power system of an all-electric ship can be characterized as an isolated microgrid system. To achieve resilient, cost-effective and privacy-preserved operation of the shipboard power system, a novel energy management strategy is introduced in this talk. Currently, a master controller is required for energy management. However, such a centralized energy management strategy suffers from numerous disadvantages. Therefore, a modified nested energy management method is proposed to preserve privacy and run the microgrid system in a distributed manner for plug-and-play operation. Furthermore, the system resilience is enhanced against energy deficiency by reserving more energy in the cloud energy storage system. This is achieved by a distributed algorithm, known as alternating direction method of multipliers (ADMM), to obtain the solution of an optimization problem with contradicting objectives. Numerical studies are presented to demonstrate the benefits of proposed energy management system.

Biography:

Kexing Lai

Dr. Kexing Lai received the Bachelor of Engineering degree in electrical engineering from Central South University, Changsha, China, in 2014. He received the Doctor of Philosophy degree in electric power systems from The Ohio State University, Columbus, OH, USA, in May 2019.
His current research interests include microgrid protection, power system planning & operation, power system resilience and reliability analysis, power system economics, and applications of machine learning in power systems.