Select a calendar:
Filter May Events by Event Type:
SUNMONTUEWEDTHUFRISAT
Events for May 06, 2016
-
CS Colloquium: John Lafferty (University of Chicago) - Statistical Learning Under Communication and Shape Constraints
Fri, May 06, 2016 @ 11:00 AM - 12:15 PM
Thomas Lord Department of Computer Science
Conferences, Lectures, & Seminars
Speaker: John Lafferty, University of Chicago
Talk Title: Statistical Learning Under Communication and Shape Constraints
Series: Yahoo! Labs Machine Learning Seminar Series
Abstract: Imagine that I estimate a statistical model from data, and then want to share my model with you. But we are communicating over a resource constrained channel. By sending lots of bits, I can communicate my model accurately, with little loss in statistical risk. Sending a small number of bits will incur some excess risk. What can we say about the tradeoff between statistical risk and the communication constraints? This is a type of rate distortion and constrained minimax problem, for which we provide a sharp analysis in certain nonparametric settings. We also consider the problem of estimating a high dimensional convex function, and develop a screening procedure to identify irrelevant variables. The approach adopts on a two-stage quadratic programming algorithm that estimates a sum of one-dimensional convex functions, beating the curse of dimensionality that holds under smoothness constraints. Joint work with Yuancheng Zhu and Min Xu.
Host: Yan Liu
Location: Ronald Tutor Hall of Engineering (RTH) - 526
Audiences: Everyone Is Invited
Contact: Assistant to CS chair
-
PhD Defense - Shay Deutsch
Fri, May 06, 2016 @ 02:30 PM - 04:30 PM
Thomas Lord Department of Computer Science
University Calendar
Title: Learning the Geometric Structure of High Dimensional Data using the Tensor Voting Graph
Location: SAL 322
Time: 2:30pm - 4:30pm, May 6th, 2016
PhD Candidate: Shay Deutsch
Committee members:
Prof. Gerard Medioni (Chair)
Prof. Aiichiro Nakano
Prof. Antonio Ortega (Outside Member)
Abstract:
This study addresses a range of fundamental problems in unsupervised manifold learning. Given a set of noisy points in a high dimensional space that lie near one or more possibly intersecting smooth manifolds, different challenges include learning the local geometric structure at each point, geodesic distance estimation, and clustering. These problems are ubiquitous in unsupervised manifold learning, and many applications in computer vision as well as other scientific applications would benefit from a principled approach to these problems.
In the first part of this thesis we present a hybrid local-global method that leverages the algorithmic capabilities of the Tensor Voting framework. However, unlike Tensor Voting, which can learn complex structures reliably only locally, our method is capable of reliably inferring the global structure of complex manifolds using a unique graph construction called the Tensor Voting Graph (TVG). This graph provides an efficient tool to perform the desired global manifold learning tasks such as geodesic distance estimation and clustering on complex manifolds, thus overcoming one of one of the main limitations of Tensor Voting as a strictly local approach. Moreover, we propose to explicitly and directly resolve the ambiguities near the intersections with a novel algorithm, which uses the TVG and the positions of the points near the manifold intersections.
In the second part of this thesis we propose a new framework for manifold denoising based processing in the graph Fourier frequency domain, derived from the spectral decomposition of the discrete graph Laplacian. The suggested approach, called MFD, uses the Spectral Graph Wavelet transform in order to perform non-iterative denoising directly in the graph frequency domain. To the best of our knowledge, MFD is the first attempt to use graph signal processing tools for manifold denoising on unstructured domains. We provide theoretical justification for our Manifold Frequency Denoising approach on unstructured graphs and demonstrate that for smooth manifolds the coordinate signals also exhibit smoothness. This is first demonstrated in the case of noiseless observations, by proving that manifolds with smoother characteristics creates more energy in the lower frequencies. Moreover, it is shown that higher frequency wavelet coefficients decay in a way that depends on the smoothness properties of the manifold, which is explicitly tied to the curvature properties. We then provide an analysis for the case of noisy points and a noisy graph, establishing results which tie the noisy graph Laplacian to the noiseless graph Laplacian characteristics, induced by the smoothness manifold properties and the graph construction properties.
Finally, the last part of this research merges the Manifold Frequency Denoising and the Tensor Voting Graph methods into a uniform framework, which allows us to denoise and analyze a general class of noisy manifolds with singularities also in the presence of outliers. We demonstrate that the limitation of the Spectral Graph Wavelets in its flexibility to analyze certain classes of graph signals can be overcome for manifolds with singularities using certain graph construction and regularization methods. This allows us to take into account global smoothness characteristics without over-smoothing in the manifold discontinuations (which correspond to high frequency bands of the Spectral Graph Wavelets), and moreover is robust to a large amount of outliers.
Location: Henry Salvatori Computer Science Center (SAL) - 322
Audiences: Everyone Is Invited
Contact: Lizsl De Leon
-
NL Seminar-THE TECHKNACQ PROJECT: BUILDING PEDAGOGICALLY TUNED READING LISTS FROM TECHNICAL CORPORA
Fri, May 06, 2016 @ 03:00 PM - 04:00 PM
Information Sciences Institute
Conferences, Lectures, & Seminars
Speaker: Gully Burns, USC/ISI
Talk Title: THE TECHKNACQ PROJECT: BUILDING PEDAGOGICALLY TUNED READING LISTS FROM TECHNICAL CORPORA
Series: Natural Language Seminar
Abstract: This work is geared towards developing pedagogically-tuned information retrieval systems to help learners select the most informative documents as a reading list for a given query over a given technical corpus. This work will enable learners to understand complex subjects more quickly. I will discuss our overall methodology, our efforts to study dependency between topics within a technical corpus and improvements to evaluating topic quality. I will describe ongoing efforts to study a document's pedagogical value to the end user and future directions for this enterprise.
Biography: Gully Burns' focus is to develop pragmatic knowledge engineering systems for scientists in collaboration with experts from the field of AI. He was originally trained as a physicist at Imperial College in London before switching to do a Ph.D. in neuroscience at Oxford. He came to work at USC in 1997, developing the 'NeuroScholar' project in Larry Swanson's lab before joining the Information Sciences Institute in 2006. He is as Research Lead at ISI.
Host: Xing Shi and Kevin Knight
More Info: http://nlg.isi.edu/nl-seminar/
Location: Information Science Institute (ISI) - 11th Flr Conf Rm # 1135, Marina Del Rey
Audiences: Everyone Is Invited
Contact: Peter Zamar
Event Link: http://nlg.isi.edu/nl-seminar/
-
Sonny Astani Department Seminar
Fri, May 06, 2016 @ 03:00 PM - 03:30 PM
Sonny Astani Department of Civil and Environmental Engineering
Conferences, Lectures, & Seminars
Speaker: Meida Chan, Civil Engineering PhD student, USC
Talk Title: Development of a data acquisition-planning framework for hybrid data collection techniques to achieve blind-spots free 3D point cloud
Abstract: As-is 3D building models are valuable in many ways such as urban planning, historical building information storage, building renovation, facility management, building energy simulation and so on. Data acquisition for complete and accurate as-is 3D building reconstruction is a time consuming and labor intensive process. Establishing a data acquisition plan before or during the data acquisition process is necessary. As such, there has been extensive research on developing/advancing data acquisition planning algorithms with a single data acquisition technique. However, for buildings that have complex building structure and architectural elements, data collection process with a single data acquisition technique is not sufficient neither effective. The hypothesis behind this research study is that image-based technique (photogrammetry) and range-based technique (laser scanning) are complementary to each other and that the combination of the two techniques can improve the quality of the derived as-is 3D point cloud in terms of completeness and accuracy. As such, this research study will develop a framework that will provide an improved data acquisition process to support the creation of complete and accurate 3D models of existing buildings, while reducing the total cost of data acquisition by eliminating the need for site revisits and reworking of the data collection process.
Host: Lucio Soibelman
Location: Seeley G. Mudd Building (SGM) - 101
Audiences: Everyone Is Invited
Contact: Kaela Berry
-
Sonny Astani Department Seminar
Fri, May 06, 2016 @ 03:30 PM - 04:00 PM
Sonny Astani Department of Civil and Environmental Engineering
Conferences, Lectures, & Seminars
Speaker: Prof. Xin Wang, Research Institute of Disaster Science, Tohoku University, Japan
Talk Title: 1D Wave Propagation Analysis and Shear-Wave Velocity Extraction of Super High-Rise Buildings Based on Ambient Vibrations Measurement
Abstract: Shaking modes of super high-rise buildings are very complex. In the first part of this study, three 2-D frame models of super high-rise buildings including bending and shear deflections in each member are used to simulate shear-wave propagation within the building. Different shaking modes at the lower stories of the three models are designed, each with a different mass-and-rigidity distributions, such as: (i) all stories shaking in a shear-bending mode, (ii) the lower eight stories shaking in pure bending mode, and (iii) the fourth to eighth stories shaking in pure bending mode. The wave reflections at the boundaries of stories with different shaking modes are examined from the response waves and the impulse responses with respect to the response of the top. Because of the wave interference, it is difficult to observe the travel path directly from the response waves. However, the travel path and the reflected waves can be observed clearly from impulse responses. For the stories shaking in pure bending mode, similar to the models (ii) and (iii), because there is no inter-story shear deformation, the apparent shear-rigidity of these stories seems infinite, which leads to zero shear-wave travel time and shear-wave velocities cannot be extracted successfully. In the second part of this study, 1D vertical shear-wave propagation in two super high-rise buildings are identified using ambient vibration response recorded by a portable array. The identified shear-wave propagation from the impulse response, including the boundary conditions, is compared with the simulated ones. Attempt is made to identify the shear-wave velocities for the individual stories.
Biography: Dr. Xin Wang is an Assistant Professor of the Research Institute of Disaster Science of Tohoku University in Japan. Her research combines knowledge of seismology and civil engineering and aims to disaster prevention from earthquakes. Her research topics include building damage detection and building damage causes examination due to ground shaking during big earthquake disasters, e.g. the 2008 Wenchuan Earthquake in China, the 2011 Great East Japan Earthquake, and the 2014 Ludian Earthquake in China. She is currently studying damage from the recent 2016 Kumamoto Earthquake in Japan. Her main recent research topics are Structural Health Monitoring of super high-rise buildings, and earthquake response recording systems using smart devices. Native of China, Dr. Wang received her B.S. and M.S. degrees in Civil Engineering from the Dalian Jiaotong University and Southeast University, respectively, after which she moved to Japan and earned her Ph.D. degree from the Aichi Institute of Technology.
Host: Maria Todorovska
Location: Seeley G. Mudd Building (SGM) - 101
Audiences: Everyone Is Invited
Contact: Kaela Berry
