Select a calendar:
Filter August Events by Event Type:
Conferences, Lectures, & Seminars
Events for August
-
Electrical Engineering Seminar
Wed, Aug 08, 2018 @ 03:03 PM - 04:30 PM
Ming Hsieh Department of Electrical and Computer Engineering
Conferences, Lectures, & Seminars
Speaker: Victor O.K. Li, The University of Hong Kong, Hong Kong
Talk Title: Bayesian Deep Learning: A Hybrid Approach to Predict Air Pollution
Abstract: Air pollution has deteriorated rapidly in many metropolitan cities, such as Beijing. Since poor air quality has clear public health impacts, accurately monitoring and predicting the concentration of PM2.5 and other pollutants have become increasingly crucial. This talk presents a hybrid approach where time series decomposition and Bayesian Long Short-Term Memory (BLSTM) are combined as a framework for air pollution forecast, based on historical data of air quality, meteorology and traffic in Beijing. LSTM has been proven to achieve state-of-the-art performance in many time series prediction applications due to its capability of memorizing long term sequential correlations. In addition, the model uncertainty estimates generated by Bayesian methods may reduce overfitting, improving the accuracy of the prediction. In our experiment, deseasonalized features are fed into BLSTM to predict the air pollution in the next 48 hours of each monitoring station in Beijing. Results show that the BLSTM framework outperforms the baseline models including SVR, STL, ARIMA, and traditional LSTM with dropout regularization.
Biography: Victor O.K. Li received SB, SM, EE and ScD degrees in Electrical Engineering and Computer Science from MIT. Prof. Li is Chair of Information Engineering and Cheng Yu-Tung Professor in Sustainable Development at the Department of Electrical & Electronic Engineering (EEE) at the University of Hong Kong. He is the Director of the HKU-Cambridge Clean Energy and Environment Research Platform, an interdisciplinary collaboration with Cambridge. He was the Head of EEE, Assoc. Dean (Research) of Engineering and Managing Director of Versitech Ltd. He serves on the board of Sunevision Holdings Ltd., listed on the Hong Kong Stock Exchange and co-founded Fano Labs Ltd., an artificial intelligence (AI) company with his PhD student. Previously, he was Professor of Electrical Engineering at the University of Southern California (USC), Los Angeles, California, USA, and Director of the USC Communication Sciences Institute. His research interests include big data, AI, optimization techniques, and interdisciplinary clean energy and environment studies. In Jan 2018, he was awarded a USD 6.3M RGC Theme-based Research Project to develop deep learning techniques for personalized and smart air pollution monitoring and health management. Sought by government, industry, and academic organizations, he has lectured and consulted extensively internationally. He has received numerous awards, including the PRC Ministry of Education Changjiang Chair Professorship at Tsinghua University, the UK Royal Academy of Engineering Senior Visiting Fellowship in Communications, the Croucher Foundation Senior Research Fellowship, and the Order of the Bronze Bauhinia Star, Government of the HKSAR. He is a Fellow of the Hong Kong Academy of Engineering Sciences, the IEEE, the IAE, and the HKIE. He can be contacted at vli@eee.hku.hk.
Host: C.-C. Jay Kuo
More Information: Victor Li Seminar Announcement.pdf
Location: Hughes Aircraft Electrical Engineering Center (EEB) - 248
Audiences: Everyone Is Invited
Contact: Gloria Halfacre
This event is open to all eligible individuals. USC Viterbi operates all of its activities consistent with the University's Notice of Non-Discrimination. Eligibility is not determined based on race, sex, ethnicity, sexual orientation, or any other prohibited factor. -
CSC@USC/CommNetS-MHI Seminar Series
Mon, Aug 27, 2018 @ 02:00 PM - 03:30 PM
Ming Hsieh Department of Electrical and Computer Engineering
Conferences, Lectures, & Seminars
Speaker: Jason Lee, University of Southern California
Talk Title: Towards Theoretical Understanding of Over-Parametrization in Deep Learning
Series: Fall 2018 Joint CSC@USC/CommNetS-MHI Seminar Series
Abstract: We provide new theoretical insights on why over-parametrization is effective in learning neural networks. For a k hidden node shallow network with quadratic activation and n training data points, we show that as long as k >= sqrt(2n) over-parametrization enables local search algorithms to find a globally optimal solution for general smooth and convex loss functions. Further, despite that the number of parameters may exceed the sample size, we show that with weight decay, the solution also generalizes well.
Next, we analyze the implicit regularization effects of various optimization algorithms. 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 is a joint work with Suriya Gunasekar, Mor Shpigel, Daniel Soudry, Nati Srebro, and Simon Du.
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: Mihailo Jovanovic, mihailo@usc.edu
Location: Hughes Aircraft Electrical Engineering Center (EEB) - 132
Audiences: Everyone Is Invited
Contact: Gerrielyn Ramos
This event is open to all eligible individuals. USC Viterbi operates all of its activities consistent with the University's Notice of Non-Discrimination. Eligibility is not determined based on race, sex, ethnicity, sexual orientation, or any other prohibited factor.
