Select a calendar:
Filter October Events by Event Type:
SUNMONTUEWEDTHUFRISAT
Events for October 14, 2016
-
AI Seminar-Matrix Completion, Saddlepoints, and Gradient Descent
Fri, Oct 14, 2016 @ 11:00 AM - 12:00 PM
Information Sciences Institute
Conferences, Lectures, & Seminars
Speaker: Jason Lee , USC
Talk Title: Matrix Completion, Saddlepoints, and Gradient Descent
Series: Artificial Intelligence Seminar
Abstract: Matrix completion is a fundamental machine learning problem with wide applications in collaborative filtering and recommender systems. Typically, matrix completion are solved by non-convex optimization procedures, which are empirically extremely successful. We prove that the symmetric matrix completion problem has no spurious local minima, meaning all local minima are also global. Thus the matrix completion objective has only saddlepoints an global minima.
Next, we show that saddlepoints are easy to avoid for even Gradient Descent -- arguably the simplest optimization procedure. We prove that with probability 1, randomly initialized Gradient Descent converges to a local minimizer.
Biography: Jason Lee is an assistant professor in Data Sciences and Operations at the University of Southern California. Prior to that, he was a postdoctoral researcher at UC Berkeley working with Michael Jordan. Jason received his PhD at Stanford University advised by Trevor Hastie and Jonathan Taylor. His research interests are in statistics, machine learning, and optimization. Lately, he has worked on high dimensional statistical inference, analysis of non-convex optimization algorithms, and theory for deep learning.
Host: Emilio Ferrara
Location: Information Science Institute (ISI) - 6th Floor -CR # 689; ISI-Marina del Rey
Audiences: Everyone Is Invited
Contact: Peter Zamar
-
Careers in Data Science and Data Engineering Infosession
Fri, Oct 14, 2016 @ 11:00 AM - 01:00 PM
Thomas Lord Department of Computer Science
Workshops & Infosessions
With the explosion of raw data being produced in recent years, data scientists and engineers work on topics covering almost every aspect of life we can think of. Top companies like Facebook, Twitter, Netflix, The New York Times, Merck, Memorial Sloan Kettering Cancer Center, and the Broad Institute are hiring students from quantitative fields to help them glean insights from the terabytes of data that they collect every day. While the amount of data these companies are producing and storing is growing exponentially, there is a severe shortage of top talent to organize, store, and analyze these data and extract valuable insights.
In this seminar, Judit Lantos will discuss her transition from a career in research to becoming a Data Engineer and Program Director at the Insight Data Engineering Fellowship program, where she helps graduating students from quantitative fields make the transition themselves. She will also give a high-level overview of the state of the field of data engineering and industry-wide trends, as well as the tools and skills that top data-driven companies are looking for when hiring data scientists and engineers. The session will include time for Q&A, as well as advice for those interested in transitioning to careers based on Insight's experience working with 600+ alumni.
Learn more at: insightdatascience.com and insightdataengineering.com
Location: Seeley G. Mudd Building (SGM) - 101
Audiences: Everyone Is Invited
Contact: Ryan Rozan
-
W.V.T. Rusch Engineering Honors Program Colloquium
Fri, Oct 14, 2016 @ 01:00 PM - 01:50 PM
USC Viterbi School of Engineering
University Calendar
Join us for a presentation by Professor Rana Adhikari, Professor of Physics, Caltech, titled "Listening to the Thunder of Gravity in the Cosmos".
Location: Henry Salvatori Computer Science Center (SAL) - 101
Audiences: Everyone Is Invited
Contact: Julie Phaneuf
-
Biomedical Engineering Speakers
Fri, Oct 14, 2016 @ 02:00 PM - 04:00 PM
Alfred E. Mann Department of Biomedical Engineering
Conferences, Lectures, & Seminars
Speaker: Danny J. Wang, PhD, Professor of Neurology, USC Institute for Neuroimaging & Informatics
Talk Title: Noncontrast Perfusion MRI and Dynamic MR Angiography
Series: Department of Biomedical Engineering: Systems Cellular-Molecular Bioengineering Distinguished Speaker Series
Abstract: Arterial spin labeling (ASL) is an emerging noninvasive MRI technique to measure cerebral blood flow (CBF) by utilizing magnetically labeled arterial blood water as an endogenous tracer. While in the past ASL has been limited by the low signal-to-noise ratio (SNR), recent technical development at high and ultrahigh magnetic fields, array receive coils and more efficient spin labeling and image acquisition methods has brought ASL to the frontier of both basic and clinical neuroscience. In this talk I will review the latest technical development of ASL and representative clinical applications in neurologic disorders such as stroke and arteriovenous malformation (AVM). Another new ASL technique is 4D non-contrast dynamic MR angiography (dMRA) that is able to visualize and quantify the dynamic flow pattern of labeled blood through the cerebrovasculature with millisecond temporal resolution and millimeter spatial resolution. I will showcase representative clinical applications of 4D dMRA, and discuss a novel method for assessing cerebral vascular compliance or arterial stiffness.
Host: Brent Liu, PhD
Location: Olin Hall of Engineering (OHE) - 100B
Audiences: Everyone Is Invited
Contact: Mischalgrace Diasanta
-
Ming Hsieh Institute Seminar Series on Integrated Systems
Fri, Oct 14, 2016 @ 02:00 PM - 04:00 PM
Ming Hsieh Department of Electrical and Computer Engineering
Conferences, Lectures, & Seminars
Speaker: Stacy Ho, Deputy Director, Analog Circuit Design, MediaTek, Inc.
Talk Title: Recent Advances in ADCs for Mobile Applications: 4G to 5G
Host: Prof. Hossein Hashemi, Prof. Mike Chen, and Prof. Mahta Moghaddam
More Information: MHI Seminar Series IS - Stacy_Ho_Flyer.pdf
Location: Hughes Aircraft Electrical Engineering Center (EEB) - 248
Audiences: Everyone Is Invited
Contact: Jenny Lin
-
Astani Civil and Environmental Engineering Ph.D. Seminar
Fri, Oct 14, 2016 @ 03:00 PM - 04:00 PM
Sonny Astani Department of Civil and Environmental Engineering
Conferences, Lectures, & Seminars
Speaker: Christopher Morrow and Ryan Gustafson, Astani CEE Ph.D. Candidates
Talk Title: See Attachment
Location: John Stauffer Science Lecture Hall (SLH) - 102
Audiences: Everyone Is Invited
Contact: Evangeline Reyes
-
Astani Civil and Environmental Engineering Ph.D. Seminar
Fri, Oct 14, 2016 @ 03:00 PM - 04:00 PM
Sonny Astani Department of Civil and Environmental Engineering
Conferences, Lectures, & Seminars
Speaker: Christopher Morrow and Ryan Gustafson, Astani CEE Ph.D. Candidates
Abstract: Towards Sustainable Wastewater Reuse: Salinity Control in Osmotic Membrane Bioreactors
By: Christopher Morrow
Alongside climate change comes an increase in drought duration and severity. In response, low energy wastewater recycling methods are becoming increasingly attractive. Our group has developed an osmotic membrane bioreactor system (OMBR) with a low energy Membrane Distillation (MD) re-concentration step driven by waste heat for potable wastewater reuse applications. A membrane bioreactor with a submerged forward osmosis (FO) module subsystem has been tested for high-strength domestic wastewater treatment. In the FO process, relatively pure water is transported from the mixed liquor into the draw solution; the mixed liquor suspended solids become concentrated and the draw solution becomes diluted. At the same time, a small amount of salt from the draw solution diffuses across the membrane from the draw solution into the mixed liquor. Consequently, bioreactor salinity increases. This reverse salt diffusion (RSD) can affect biological processes, particularly at higher concentrations. In this presentation, RSD from two membrane configurations will be evaluated; the first is a membrane cassette submerged in the bioreactor and the second is a side-stream process external to the bioreactor. In the submerged configuration, the draw solution is recirculated through a frame skinned with FO membrane and in the external configuration, both the feed and draw solutions have crossflow along the membrane selective and structural layers, respectively. Two advantages of external operation are: 1) crossflow on the feed side scours foulants away from the membrane, and 2) increasing crossflow velocity (CFV) on the draw side reduces dilutive concentration polarization. These result in increased water flux, but may also change RSD, which alters the composition of the foulant layer and/or the bioreactor salinity. Bench-scale FO experiments to determine the specific reverse salt flux (SRSF) with each configuration were carried out. The bench-scale results were used as inputs to initialize a model and predict the steady-state salt concentration of the pilot-scale bioreactor. Further analysis was carried out to examine the affect of changing the solids retention time (SRT), membrane area, and reactor volume to optimize system performance and maintain low salinity OMBRs.
TITLE: Waste-heat-driven membrane distillation (MD) and long-term hydrophobicity of MD membranes
By: Ryan Gustafson
Abstract:
Global concern regarding water scarcity, climate change, and environmental health have resulted in increased interest in new water treatment technologies, particularly technologies for water reuse and desalination applications with low electrical energy requirements. Increased interest in water reuse and environmental health have resulted in more stringent regulatory requirements for producers of industrial and municipal wastewaters. One technology poised to address these concerns is MD. MD is a thermally driven separation process that is used to distill water from an impaired feed water source using a hydrophobic membrane. Maintaining the hydrophobicity of the MD membrane is vital to maintaining the characteristic high rejection of non-volatile contaminants that is key to the application of MD to water treatment. It is though that long-term exposure of MD membranes to flowing water at high temperature and high salinity will result in reduced membrane hydrophobicity over time, but this has not been proven. Another key aspect of MD technology for its application to water treatment is its ability to be driven by low-grade waste-heat. While most MD researchers assume that low-grade waste-heat will be available and easily transferrable to the MD module, few have demonstrated successful operation of waste-heat-driven MD systems. Further, the small amount of research available on these systems lacks detailed analysis of the impacts of waste heat source variability on water production. Finally, an analysis of system-wide heat transfer behavior and the impact of different system configurations on water production in WHD-MD systems is not available in existing studies. In my presentation, I will discuss
Location: John Stauffer Science Lecture Hall (SLH) - 102
Audiences: Everyone Is Invited
Contact: Evangeline Reyes
-
EMNLP PRACTICE TALK: UNDERSTANDING NEURAL MACHINE TRANSLATION: LENGTH CONTROL AND SYNTACTIC STRUCTURE
Fri, Oct 14, 2016 @ 03:00 PM - 04:00 PM
Information Sciences Institute
Conferences, Lectures, & Seminars
Speaker: Xing Shi, USC/ISI
Talk Title: EMNLP PRACTICE TALK: UNDERSTANDING NEURAL MACHINE TRANSLATION: LENGTH CONTROL AND SYNTACTIC STRUCTURE
Series: Natural Language Seminar
Abstract: Neural Machine Translation is powerful but we know little about the black box. We conduct the following two investigations to gain a better understanding: First, we investigate how neural, encoder-decoder translation systems output target strings of appropriate lengths, finding that a collection of hidden units learns to explicitly implement this functionality. Second, we investigate whether a neural, encoderdecoder translation system learns syntactic information on the source side as a by-product of training. We propose two methods to detect whether the encoder has learned local and global source syntax. A fine-grained analysis of the syntactic structure learned by the encoder reveals which kinds of syntax are learned and which are missing.
Biography: Xing Shi is a PhD student at ISI working with Prof. Kevin Knight.
Host: Kevin Knight
More Info: http://nlg.isi.edu/nl-seminar/
Location: Information Science Institute (ISI) - 6th Floor -CR # 689; ISI-Marina del Rey
Audiences: Everyone Is Invited
Contact: Peter Zamar
Event Link: http://nlg.isi.edu/nl-seminar/
