Colloquia Series-Jason Hartline Series-Jason Hartline2018-10-26T05:00:00Z11:00 a.m.Lopata Hall, Room 101<p style="text-align: center;"><strong>Data Science and Mechanism Design</strong></p><p><strong>Abstract</strong></p><p><strong></strong>Computer systems have become the primary mediator of social and economic interactions.  A defining aspect of such systems is that the participants have preferences over system outcomes and will manipulate their behavior to obtain outcomes they prefer.  Such manipulation interferes with data-driven methods for designing and testing system improvements.  A standard approach to resolve this interference is to infer preferences from behavioral data and employ the inferred preferences to evaluate novel system designs.</p><p>In this talk I will describe a method for estimating and comparing the performance of novel systems directly from behavioral data from the original system.  This approach skips the step of estimating preferences and is more accurate.  Estimation accuracy can be further improved by augmenting the original system; its accuracy then compares favorably with ideal controlled experiments, a.k.a., A/B testing, which are often infeasible.  A motivating example will be the paradigmatic problem of designing an auction for the sale of advertisements on an Internet search engine. </p><p><strong>Biography</strong></p><p>Jason Hartline is an associate professor of computer science at Northwestern University. His research introduces design and analysis methodologies from computer science to understand and improve outcomes of economic systems. Optimal behavior and outcomes in complex environments are complex and, therefore, should not be expected; instead, the theory of approximation can show that simple and natural behaviors are approximately optimal in complex environments. This approach is applied to auction theory and mechanism design in his graduate textbook Mechanism Design and Approximation which is under preparation ( Prof. Hartline received his Ph.D. in 2003 from the University of Washington under the supervision of Anna Karlin. He was a postdoctoral fellow at Carnegie Mellon University under the supervision of Avrim Blum; and subsequently a researcher at Microsoft Research in Silicon Valley. He joined Northwestern University in 2008.<br/></p>Eugene Vorobeychik
Colloquia Series-Joshua Levine Series-Joshua Levine2018-11-02T05:00:00Z11:00 a.m.Lopata Hall, Room 101<p style="text-align: center;"><strong>Visualizing Scalar Data with Computational Topology and Machine Learning</strong></p><p><strong>Abstract</strong></p>In this talk, I will discuss and demonstrate two visualization projects.  The first is recent work with collaborators at UPMC Sorbonne on the creation of the Topological ToolKit (TTK), an open source software platform for topological data analysis of piecewise linear scalar fields.  TTK is built on top of VTK and ParaView, and provides access points for developers, end users, and researchers across a wide of range of experience levels.  Two key advantages of TTK are that it (1) provides a unified platform for topological analysis, allowing a modular approach to analyzing data under a consistent set of mathematical abstractions and (2) offers an efficient triangulation data structure that caches queries for repeated use.  TTK is available on the web at<div><br/><p>In the second part of my talk, I will describe a new project with collaborators at Vanderbilt University that studies how generative models can be used to model the process of volume rendering scalar fields.  We construct a generative adversarial network that learns the mapping from volume rendering parameters, such as viewpoint and transfer function, to the rendered image.  In doing so, we can analyze the volume itself and provide new mechanisms for guiding the user in transfer function editing and exploring the space of possible images that can be volume rendered.  Both our training process and applications are available on the web at</p><p><strong>Biography</strong><br/></p><p>Joshua A. Levine ( is an assistant professor in the Department of Computer Science at University of Arizona. Prior to starting at Arizona in 2016, he was an assistant professor at Clemson University from 2012 to 2016, and before that a postdoctoral research associate at the University of Utah's SCI Institute from 2009 to 2012. He received his PhD in Computer Science from The Ohio State University in 2009 after completing BS degrees in Computer Engineering and Mathematics in 2003 and an MS in Computer Science in 2004 from Case Western Reserve University. His research interests include visualization, geometric modeling, topological analysis, mesh generation, vector fields, performance analysis, and computer graphics.<br/></p></div>Alvitta Ottley
Colloquia Series-Alec Koppel Series-Alec Koppel2018-11-16T06:00:00Z11:00 a.m.Lopata Hall, Room 101<p style="text-align: center;"><strong>Machine Learning for Time Series via Optimally Compressed Bayesian Methods</strong></p><p><strong>Abstract</strong></p><p><strong></strong>Machine learning is ubiquitous across many areas of science and engineering. Today's learning pipeline involves storing big data in the cloud where it is used to train deep networks, which has seen impressive successes in vision and speech recognition. However, this framework fails to address application domains where data perpetually changes such as field robotics and econometrics. In this talk, we propose methods based on Bayesian statistics that can stably adapt in the face of changing data. Unfortunately, Bayesian methods classically suffer from the curse of dimensionality: their model complexity is proportionate to the time index, and so popular perception is they are inapplicable to streaming data scenarios. We survey our recent efforts to upend this perception through the introduction of online compression rules that nearly preserve optimality while ensuring the model complexity is at-worst finite. Specifically, we'll discuss compression rules for kernel regression/classification, their extensions in risk-aware learning and reinforcement learning, as well as Gaussian Process regression. These methods pave the way for new systems that may autonomously and accurately adapt.</p><p><strong>Biography</strong></p><p>Alec Koppel began as a Research Scientist at the U.S. Army Research Laboratory in the Computational and Information Sciences Directorate in September of 2017. He completed his Master's degree in Statistics and Doctorate in Electrical and Systems Engineering, both at the University of Pennsylvania (Penn) in August of 2017. He is also a participant in the Science, Mathematics, and Research for Transformation (SMART) Scholarship Program sponsored by the American Society of Engineering Education. Before coming to Penn, he completed his Master's degree in Systems Science and Mathematics and Bachelor's Degree in Mathematics, both at Washington University in St. Louis (WashU), Missouri. His research interests are in the areas of signal processing, optimization and learning theory. His current work focuses on optimization and learning methods for streaming data applications, with an emphasis on problems arising in autonomous systems. He co-authored a paper selected as a Best Paper Finalist at the 2017 IEEE Asilomar Conference on Signals, Systems, and Computers.<br/></p>Ron Cytron
No Classes (Thanksgiving Break) Classes (Thanksgiving Break)2018-11-21T06:00:00Z
No Classes (Thanksgiving Break) Classes (Thanksgiving Break)2018-11-22T06:00:00Z
No Classes (Thanksgiving Break) Classes (Thanksgiving Break)2018-11-23T06:00:00Z
Last Day of Fall 2018 Classes Day of Fall 2018 Classes2018-12-07T06:00:00Z
Reading Day (No Classes) Day (No Classes)2018-12-10T06:00:00Z<p>No lectures, laboratories or exams for School of Engineering & Applied Science.​</p>Engineering Student Services, 314-935-6100