Select a calendar:
Filter November Events by Event Type:
SUNMONTUEWEDTHUFRISAT
Events for the 3rd week of November
-
Remarkable Trajectory Lecture by Gerard Medioni
Mon, Nov 11, 2019 @ 04:00 PM - 05:30 PM
Thomas Lord Department of Computer Science
Conferences, Lectures, & Seminars
Speaker: Prof. Gerard Medioni, USC
Talk Title: 40+ Years of Computer Vision at USC
Series: Remarkable Trajectory Lecture Series
Host: Computer Science Department
Location: University Club of USC, Scriptorium Room
Audiences: Everyone Is Invited
Contact: Computer Science Department
-
CAIS Seminar: Sheldon H. Jacobson (University of Illinois) - Creating a Transparent Environment for Political Redistricting
Wed, Nov 13, 2019 @ 02:00 PM - 03:00 PM
Thomas Lord Department of Computer Science
Conferences, Lectures, & Seminars
Speaker: Dr. Sheldon H. Jacobson, University of Illinois
Talk Title: Creating a Transparent Environment for Political Redistricting
Series: USC Center for Artificial Intelligence in Society (CAIS) Seminar Series
Abstract: Political redistricting is a multi-criteria problem with conflicting objectives (based on metrics like compactness, population balance, and efficiency gaps, among others). Many of these metrics have received significant attention, though they remain controversial as to which such metrics are best suited to define fair district maps. This research uses a multi-objective optimization approach to reveal obstacles in defining fair district maps. The results obtained challenge a number of common perceptions of redistricting, suggesting that defining fair maps may not only be extremely difficult, but also, simply unrealistic.
This lecture satisfies requirements for CSCI 591: Research Colloquium. Please note, due to limited capacity in OHE 136, seats will be first come first serve.
Biography: Sheldon H. Jacobson is a Founder Professor of Computer Science at the University of Illinois. He has a B.Sc. and M.Sc. (both in Mathematics) from McGill University, and a M.S. and Ph.D. (both in Operations Research) from Cornell University. From 2012-2014, he was on leave from the University of Illinois, serving as a Program Director at the National Science Foundation. His research interests span theory and practice, covering decision-making under uncertainty and optimization-based artificial intelligence, with applications in aviation security, public policy, public health, and sports. He has been recognized by numerous awards, including a Guggenheim Fellowship from the John Simon Guggenheim Memorial Foundation. He is a fellow of both IISE and INFORMS.
Host: USC Center for Artificial Intelligence in Society (CAIS)
Location: Olin Hall of Engineering (OHE) - 136
Audiences: Everyone Is Invited
Contact: Computer Science Department
-
PhD Thesis Proposal - Michael Tsang
Thu, Nov 14, 2019 @ 10:00 AM - 11:30 PM
Thomas Lord Department of Computer Science
University Calendar
Title: Interpretable Machine Learning Models via Feature Interaction Discovery
Date/Time: Thursday, November 14th 10-11:30am
Location: SAL 322
Candidate: Michael Tsang
Committee: Prof. Yan Liu (adviser), Prof. Joseph Lim, Prof. Maja Mataric, Prof. Emily Putnam Hornstein, Prof. Xiang Ren
The impact of machine learning prediction models has created a growing need for us to understand why they make their predictions. The interpretation of these models is important to reveal their fundamental behavior, to obtain scientific insights into data, and to help us trust automatic predictions. In this thesis proposal, we advance these directions via the problem of feature interaction discovery. We develop a way to interpret the feature interactions in feedforward neural networks by tracing their learned weights. We follow-up on this method and develop a way of learning transparent neural networks. Lastly, we investigate applications of this work on interpreting black-box models beyond feedforward neural networks, such as image/text classifiers and recommender systems. Throughout this presentation, we will explain the physical meaning and practical importance of our feature interaction interpretations.Location: Henry Salvatori Computer Science Center (SAL) - 322
Audiences: Everyone Is Invited
Contact: Lizsl De Leon
-
Theory Lunch
Thu, Nov 14, 2019 @ 12:15 PM - 02:00 PM
Thomas Lord Department of Computer Science
Conferences, Lectures, & Seminars
Speaker: Mengxiao Zhang, CS PhD Student
Talk Title: Gradient Descent Provably Optimizes Over-Parameterized Neural Networks
Abstract: This talk is on the paper "Gradient Descent Provably Optimizes Over-Parameterized Neural Networks," which is about how techniques like gradient descent have zero training loss even for objective functions that are non-convex and non-smooth.
Host: Shaddin Dughmi
Location: Henry Salvatori Computer Science Center (SAL) - 213
Audiences: Everyone Is Invited
Contact: Cherie Carter
-
CS Colloquium: Bryan Perozzi (Google AI) - Machine Learning on Graphs
Thu, Nov 14, 2019 @ 03:30 PM - 04:50 PM
Thomas Lord Department of Computer Science
Conferences, Lectures, & Seminars
Speaker: Bryan Perozzi, Google AI
Talk Title: Machine Learning on Graphs
Series: Computer Science Colloquium
Abstract: Machine Learning on Graphs (also known as Relational Learning, or Graph-Based Machine Learning) is a branch of ML which focuses on problems where the data items (nodes) contain discrete relationships (edges) between themselves (usually in addition to traditional real-valued feature vectors). The structure of these links between unlabelled data items can be leveraged for both semi-supervised learning and unsupervised learning algorithms.
In this talk, I will provide an overview of the area, and some recent results from our team in clustering and representation learning. When appropriate, I will try to motivate our research with examples of real world problems.
This lecture satisfies requirements for CSCI 591: Research Colloquium.
Biography: Bryan Perozzi is a Senior Research Scientist in Google AI's Algorithms and Optimization group, where he routinely analyzes some of the world's largest (and perhaps most interesting) graphs. Bryan's research focuses on developing techniques for learning expressive representations of relational data with neural networks. These scalable algorithms are useful for prediction tasks (classification/regression), pattern discovery, and anomaly detection in large networked data sets.
Bryan is an author of 20+ peer-reviewed papers at leading conferences in machine learning and data mining (such as ICML, NeurIPS, KDD, and WWW). His doctoral work on learning network representations was awarded the 2017 KDD Dissertation Award. Bryan received his Ph.D. in Computer Science from Stony Brook University in 2016, and his M.S. from the Johns Hopkins University in 2011.
Host: Sami Abu-El-Haija
Location: Henry Salvatori Computer Science Center (SAL) - 101
Audiences: Everyone Is Invited
Contact: Computer Science Department
