Select a calendar:
Filter February Events by Event Type:
SUNMONTUEWEDTHUFRISAT
Conferences, Lectures, & Seminars
Events for February
-
CAIS Seminar: Tanya Berger-Wolf (University of Illinois at Chicago) - Computational Behavioral Ecology: Animals as Mobile Social Users
Wed, Feb 06, 2019 @ 04:00 PM - 05:00 PM
Thomas Lord Department of Computer Science
Conferences, Lectures, & Seminars
Speaker: Dr. Tanya Berger-Wolf, University of Illinois at Chicago
Talk Title: Computational Behavioral Ecology: Animals as Mobile Social Users
Series: USC Center for Artificial Intelligence in Society (CAIS) Seminar Series
Abstract: New data collection technology, such as GPS, high definition cameras, UAVs, genotyping, and crowdsourcing, are generating data about wild populations that are orders of magnitude richer than any previously collected. Unfortunately, in this domain as in many others, our ability to analyze data lags substantially behind our ability to collect it. In this talk, Dr. Berger-Wolf will show how computational approaches can be part of every stage of the scientific process of understanding animal sociality, from intelligent data collection (crowdsourcing photographs and identifying individual animals from photographs by stripes and spots) to hypothesis formulation (by designing a novel computational framework for analysis of dynamic social networks), and provide scientific insight into collective behavior of zebras, baboons, and other social animals, including humans.
This lecture satisfies requirements for CSCI 591: Research Colloquium
Biography: Dr. Tanya Berger-Wolf is a Professor of Computer Science at the University of Illinois at Chicago, where she heads the Computational Population Biology Lab. As a computational ecologist, her research is at the unique intersection of computer science, wildlife biology, and social sciences. Berger-Wolf is also a director and co-founder of the AI for conservation non-profit Wild Me, home of the Wildbook project, which recently enabled the first-of-its-kind complete species census of the endangered Grevy's zebra, using photographs taken by ordinary citizens in Kenya.
Host: Milind Tambe
Location: James H. Zumberge Hall Of Science (ZHS) - 252
Audiences: Everyone Is Invited
Contact: Computer Science Department
-
CS Colloquium: Bonnie Lei (Microsoft) - AI for Earth: Tackling Global Environmental Challenges
Thu, Feb 07, 2019 @ 04:00 PM - 05:20 PM
Thomas Lord Department of Computer Science
Conferences, Lectures, & Seminars
Speaker: Bonnie Lei, Microsoft
Talk Title: AI for Earth: Tackling Global Environmental Challenges
Series: Computer Science Colloquium
Abstract: Monitoring Earth's conditions, including our air, water, land, and the well-being of our wildlife helps us better understand the dire environmental challenges our planet is currently facing. But we need the power of technological approaches such as AI to convert this vast amount of data into implementable insights fast enough to better manage our natural resources.
This inspired Microsoft to launch the AI for Earth program in 2017, committing $50 million over the next 5 years to help researchers and innovators leverage artificial intelligence technology for environmental solutions in the areas of climate, water, agriculture, and biodiversity conservation. Bonnie will share several examples of how AI for Earth and its partners have done this during the program's first year, including:
- USC Center for AI and Society's PAWS algorithm: integrates machine learning to predict poachers' behavior and plan the most effective patrol routes for rangers in protected areas
- Wild Me: Using computer vision and deep learning algorithms to scan and identify individual animals from scientific data and social media
- Project Premonition: using cloud-scale genomics and machine learning algorithms to better understand biodiversity from blood-sucking insects
This lecture satisfies requirements for CSCI 591: Research Colloquium.
Biography: Bonnie Lei is currently the project manager for Microsoft's AI for Earth program, where she leads its strategic partnerships and grants program. Previously, she traveled the globe as an environmental scientist and conservationist. She helped start the marine program for Wildlife Conservation Society in Myanmar, discovered a new sea slug species in the Caribbean, and researched climate adaptation of endangered penguins in South Africa. She has degrees in biology and economics & business from Harvard and Tsinghua Universities.
Host: Computer Science Department
Location: Henry Salvatori Computer Science Center (SAL) - 101
Audiences: Everyone Is Invited
Contact: Computer Science Department
-
CS Colloquium: Sangeetha Abdu Jyothi (University of Illinois at Urbana-Champaign) - Automated Resource Management in Large-Scale Networked Systems
Tue, Feb 12, 2019 @ 11:00 AM - 12:00 PM
Thomas Lord Department of Computer Science
Conferences, Lectures, & Seminars
Speaker: Sangeetha Abdu Jyothi, University of Illinois at Urbana-Champaign
Talk Title: Automated Resource Management in Large-Scale Networked Systems
Series: CS Colloquium
Abstract: Internet applications rely on large-scale networked environments such as the cloud for their backend support. In these multi-tenanted environments, various stakeholders have diverse goals. The objective of the infrastructure provider is to increase revenue by utilizing the resources efficiently. Applications, on the other hand, want to meet their performance requirements at minimal cost. However, estimating the exact amount of resources required to meet the application needs is a difficult task, even for expert users. Easy workarounds employed for tackling this problem, such as resource over-provisioning, negatively impact the goals of the provider, applications, or both.
In this talk, I will discuss the design of application-aware self-optimizing systems through automated resource management that helps meet the varied goals of the provider and applications in large-scale networked environments. The key steps in closed-loop resource management include learning of application resource needs, efficient scheduling of resources, and adaptation to variations in real time. I will describe how I apply this high-level approach in two distinct environments using (a) Morpheus in enterprise clusters, and (b) Patronus in cellular provider networks with geo-distributed micro data centers. I will also touch upon my related work in application-specific context at the intersection of network scheduling and deep learning. I will conclude with my vision for self-optimizing systems including fully automated clouds and an elastic geo-distributed platform for thousands of micro data centers.
This lecture satisfies requirements for CSCI 591: Research Colloquium.
Biography: Sangeetha Abdu Jyothi is a Ph.D. candidate at the University of Illinois at Urbana-Champaign. Her research interests lie in the areas of computer networking and systems with a focus on building application-aware self-optimizing systems through automated resource management. She is a winner of the Facebook Graduate Fellowship (2017-2019) and the Mavis Future Faculty Fellowship (2017-2018). She was invited to attend the Rising Stars in EECS workshop at MIT (2018).
Website: http://abdujyo2.web.engr.illinois.edu
Host: Barath Raghavan
Location: Olin Hall of Engineering (OHE) - 100 D
Audiences: Everyone Is Invited
Contact: Assistant to CS chair
-
CS Colloquium: Skip Rizzo (USC) - The Birth of Intelligent Virtual Human Agents in Clinical Healthcare
Wed, Feb 13, 2019 @ 07:30 PM - 09:00 PM
Thomas Lord Department of Computer Science
Conferences, Lectures, & Seminars
Speaker: Skip Rizzo, University of Southern California Institute for Creative Technologies
Talk Title: The Birth of Intelligent Virtual Human Agents in Clinical Healthcare
Series: Computer Science Colloquium
Abstract: Since the mid-1990s, a significant scientific literature has evolved regarding the mental/physical health outcomes from the use of what we now refer to as Clinical Virtual Reality (VR). While the preponderance of clinical work with VR has focused on building immersive virtual worlds for treating anxiety disorders with exposure therapy, providing distracting immersive experiences for acute pain management, and supporting physical rehabilitation with game-based interactive content, there are other emerging areas that have extended the impact of VR in healthcare. One such area involves the evolution of conversational virtual human (VH) agents. This has been driven by seminal research and development leading to the creation of highly interactive, artificially intelligent and natural language capable VHs that can engage real human users in a credible fashion. No longer at the level of a prop to add context or minimal faux interaction in a virtual world, VH representations can now be designed to perceive and act in a 3D virtual world, engage in face-to-face spoken dialogues with real users, and in some cases, can exhibit human-like emotional reactions. This presentation will provide a brief rationale and overview of research that has shown the benefits derived from the use of virtual humans in healthcare applications. Research will be detailed reporting positive outcomes from studies using VHs in the role of virtual patients for training novice clinicians, as job interview/social skill trainers for persons on the autism spectrum, and as online healthcare support agents with university students and military Veterans. The computational capacity now exists to deliver similar VH interactions by way of mobile device technology. This capability can support the "anywhere/anytime" availability of VH characters as agents for engaging users with healthcare information and could provide opportunities for improving access to care and emotional support for a wide range of wellness and clinical applications for a variety of populations. This work will be discussed along with a look into the future of this next major movement in Clinical VR.
RSVP: https://goo.gl/forms/TGfFn2X6h0XGQMun1
This lecture satisfies requirements for CSCI 591: Research Colloquium.
Biography: Skip Rizzo is a clinical psychologist and Director of Medical VR at the University of Southern California Institute for Creative Technologies. He is also a Research Professor with the USC Dept. of Psychiatry and School of Gerontology. Over the last 25 years, Skip has conducted research on the design, development and evaluation of Virtual Reality systems targeting the areas of clinical assessment, treatment, and rehabilitation across the domains of psychological, cognitive and motor functioning in both healthy and clinical populations. This work has focused on PTSD, TBI, Autism, ADHD, Alzheimer's disease, stroke and other clinical conditions. He has also driven an extensive research program on the use of intelligent virtual human agents for clinical training, healthcare information support, and clinical assessment. In spite of the diversity of these clinical R&D areas, the common thread that drives all of his work with digital technologies involves the study of how Virtual Reality simulations can be usefully applied to human healthcare beyond what is possible with traditional 20th Century methods.
Host: AAAI@USC
Location: James H. Zumberge Hall Of Science (ZHS) - 159
Audiences: Everyone Is Invited
Contact: Computer Science Department
-
CS Colloquium: Jason Lee (USC, Data Sciences and Operations)On the Foundations of Deep Learning: SGD, Overparametrization, and Generalization
Tue, Feb 19, 2019 @ 11:00 AM - 12:00 PM
Thomas Lord Department of Computer Science
Conferences, Lectures, & Seminars
Speaker: Jason Lee, USC, Data Sciences and Operations
Talk Title: On the Foundations of Deep Learning: SGD, Overparametrization, and Generalization
Series: CS Colloquium
Abstract: We provide new results on the effectiveness of SGD and overparametrization in deep learning.
a) SGD: We show that SGD converges to stationary points for general nonsmooth , nonconvex functions, and that stochastic subgradients can be efficiently computed via Automatic Differentiation. For smooth functions, we show that gradient descent, coordinate descent, ADMM, and many other algorithms, avoid saddle points and converge to local minimizers. For a large family of problems including matrix completion and shallow ReLU networks, this guarantees that gradient descent converges to a global minimum.
b) Overparametrization: We show that gradient descent finds global minimizers of the training loss of overparametrized deep networks in polynomial time.
c) Generalization:
For general neural networks, we establish a margin-based theory. The minimizer of the cross-entropy loss with weak regularization is a max-margin predictor, and enjoys stronger generalization guarantees as the amount of overparametrization increases.
d) Algorithmic and Implicit Regularization: We analyze the implicit regularization effects of various optimization algorithms on overparametrized networks. In particular we prove that for least squares with mirror descent, the algorithm converges to the closest solution in terms of the bregman divergence. For linearly separable classification problems, we prove that the steepest descent with respect to a norm solves SVM with respect to the same norm. For over-parametrized non-convex problems such as matrix sensing or neural net with quadratic activation, we prove that gradient descent converges to the minimum nuclear norm solution, which allows for both meaningful optimization and generalization guarantees
This lecture satisfies requirements for CSCI 591: Research Colloquium
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: Yan Liu
Location: Olin Hall of Engineering (OHE) - 132
Audiences: Everyone Is Invited
Contact: Assistant to CS chair
-
CAIS Seminar: Kristina Lerman (USC) - Friendship Paradox and Information Bias in Networks
Wed, Feb 20, 2019 @ 04:00 PM - 05:00 PM
Thomas Lord Department of Computer Science
Conferences, Lectures, & Seminars
Speaker: Dr. Kristina Lerman, University of Southern California Information Sciences Institute
Talk Title: Friendship Paradox and Information Bias in Networks
Series: USC Center for Artificial Intelligence in Society (CAIS) Seminar Series
Abstract: Individuals' decisions, from what product to buy to who to vote for, often depend on what others are doing. People, however, rarely have global information about others, but must estimate it from the local observations they make of their friends. Dr. Lerman discusses the counter-intuitive phenomena by which the structure of social networks significantly distorts the observations people make of their friends. The effects include the "friendship paradox," which states that your friends have more friends than you do, on average, and its many more surprising generalizations. As a result of these paradoxes, a trait that is globally rare may be dramatically over-represented in the local neighborhoods of many people. Friendship paradoxes may lead individuals to systematically overestimate the prevalence of a minority opinion or behavior, and may accelerate the spread of social contagions and adoption of social norms.
This lecture satisfies requirements for CSCI 591: Research Colloquium.
Biography: Kristina Lerman is a Principal Scientist at the University of Southern California Information Sciences Institute and holds a joint appointment as a Research Associate Professor in the department of Computer Science at the USC Viterbi School of Engineering. Trained as a physicist, she now applies network analysis and machine learning to problems in computational social science, including crowdsourcing, social network and social media analysis. Her recent work on modeling and understanding cognitive biases in social networks has been covered by the Washington Post, Wall Street Journal, and MIT Tech Review.
Host: Milind Tambe
Location: James H. Zumberge Hall Of Science (ZHS) - 252
Audiences: Everyone Is Invited
Contact: Computer Science Department
-
CS Colloquium: Marco Gaboardi (University at Buffalo, SUNY) Differential Privacy: Formal Verification and Applications
Thu, Feb 21, 2019 @ 11:00 AM - 12:00 PM
Thomas Lord Department of Computer Science
Conferences, Lectures, & Seminars
Speaker: Marco Gaboardi, University at Buffalo, SUNY
Talk Title: Differential Privacy: Formal Verification and Applications
Series: CS Colloquium
Abstract: A vast amount of individuals' data is collected, stored and accessed every day. These data are valuable for scientific and medical research, for decision making, etc. However, use or release of these data may be restricted by concerns for the privacy of the individuals contributing them.
Differential Privacy has been conceived to offer ways to answer statistical queries about sensitive data while providing strong provable privacy guarantees ensuring that the presence or absence of a single individual in the data has a negligible statistical effect on the query's result. In this talk I will present some formal verification techniques we developed to help programmers to certify their programs differentially private and to guarantee that their programs provide accurate answers. These techniques combine approaches based on type systems and program logics with ideas for reasoning about differential privacy using composition, sensitivity and probabilistic coupling. This combination permits fine-grained formal analyses of several basic mechanisms that are fundamental for designing practical differential privacy applications. In addition, I will present some of our results showing how to answer a large number of queries on high dimensional datasets preserving privacy, and how to perform differentially private chi-squared hypothesis testing with the same asymptotic guarantees as the traditional tests.
This lecture satisfies requirements for CSCI 591: Research Colloquium
Biography: Marco Gaboardi is an assistant professor at the University at Buffalo, SUNY, and a visiting scholar at the Simons Institute for the Theory of Computing. Prior to joining Buffalo, he was an assistant professor at the University of Dundee, Scotland. Marco received his PhD from the University of Torino, Italy, and the Institute National Polytechnique de Lorraine, France. He has been a visitor scholar at the University of Pennsylvania and at Harvard's CRCS center, and a recipient of a EU Marie Curie Fellowship. Marco's research is in programming languages, formal verification, and in differential privacy.
Host: Jyotirmoy Deshmukh
Location: Olin Hall of Engineering (OHE) - 132
Audiences: Everyone Is Invited
Contact: Assistant to CS chair
-
CS Colloquium: Rajalakshmi Nandakumar (University of Washington) - Computational Wireless Sensing at Scale
Tue, Feb 26, 2019 @ 11:00 AM - 12:00 PM
Thomas Lord Department of Computer Science
Conferences, Lectures, & Seminars
Speaker: Rajalakshmi Nandakumar, University of Washington
Talk Title: Computational Wireless Sensing at Scale
Series: CS Colloquium
Abstract: Computational wireless sensing is an exciting field of research where we use wireless signals from everyday computing devices to enable sensing. The key challenge is to enable new sensing capabilities that can be deployed at scale and have an impact in the real world.
In this talk, I will focus on the two unique approaches that I pursued to enable sensing at scale. The first is to transform existing smart devices such as smartphones into active sonar systems to enable mobile health and user interaction applications. In particular, I will talk about contactless sensing of physiological signals like breathing using off-the-shelf smartphones that can be used to detect potentially life-threatening conditions such as opioid overdoses as well as sleep apnea. The second approach is to design new low power wireless technologies that can enable IoT sensing on everyday objects on a large scale by addressing power and size constraints. Here, I will talk about our technology that achieves 3D localization and tracking of sub-centimeter sized devices that enables applications ranging from user interaction to precision agriculture.
This lecture satisfies requirements for CSCI 591: Research Colloquium
Biography: Rajalakshmi Nandakumar is a Ph.D. candidate at the Paul G. Allen School of computer science of University of Washington. Her research focuses on developing wireless sensing technologies that enable novel applications in various domains including mobile health, user interfaces and IoT networks. She developed the first contactless smartphone based sleep apnea diagnosis system that was licensed by ResMed Inc. and now used by millions of users for sleep staging. She was recognized with the Paul Baran Young Scholar award by the Marconi Society in 2018 and also named as the rising star in EECS by MIT. She has first author papers in top medical journals including Science translational medicine as well as computer science venues (CHI, SIGCOMM, SenSys, MobiCom, MobiSys). Her research was awarded multiple accolades and nominations including MobiSys 2015 best paper nominee, CHI 2016 Honorable mention award and SenSys 2018 best paper award.
Host: Ramesh Govindan
Location: Olin Hall of Engineering (OHE) - 132
Audiences: Everyone Is Invited
Contact: Assistant to CS chair
-
MASCLE Machine Learning Seminar: Jacob Abernethy (Georgia Tech) - Building Algorithms by Playing Games
Tue, Feb 26, 2019 @ 04:00 PM - 05:20 PM
Thomas Lord Department of Computer Science
Conferences, Lectures, & Seminars
Speaker: Jacob Abernethy, Georgia Institute of Technology
Talk Title: Building Algorithms by Playing Games
Series: Visa Research Machine Learning Seminar Series hosted by USC Machine Learning Center
Abstract: A very popular trick for solving certain types of optimization problems is this: write your objective as the solution of a two-player zero-sum game, endow both players with an appropriate learning algorithm, watch how the opponents compete, and extract an (approximate) solution from the actions/decisions taken by the players throughout the process. This approach is very generic and provides a natural template to produce new and interesting algorithms. I will describe this framework and show how it applies in several scenarios, and describe recent work that draws a connection to the Frank-Wolfe algorithm and Nesterov's Accelerated Gradient Descent.
This lecture satisfies requirements for CSCI 591: Research Colloquium.
Biography: Jacob Abernethy is an Assistant Professor in Computer Science at Georgia Tech. He started his faculty career in the Department of Electrical Engineering and Computer Science at the University of Michigan. In October 2011 he finished a PhD in the Division of Computer Science at the University of California at Berkeley, and then spent nearly two years as a Simons postdoctoral fellow at the CIS department at UPenn. Abernethy's primary interest is in Machine Learning, with a particular focus in sequential decision making, online learning, online algorithms and adversarial learning models.
Host: Haipeng Luo
Location: Henry Salvatori Computer Science Center (SAL) - 101
Audiences: Everyone Is Invited
Contact: Computer Science Department
