Select a calendar:
Filter April Events by Event Type:
SUNMONTUEWEDTHUFRISAT
Events for April 09, 2013
-
Nam Ma (USC), PhD Defense
Tue, Apr 09, 2013 @ 10:00 AM - 12:00 PM
Thomas Lord Department of Computer Science
University Calendar
Committee:
Aiichiro Nakano
Viktor K. Prasanna (Chairman)
Cauligi S. Raghavendra
Title: Scalable Exact Inference in Probabilistic Graphical Models on Multi-core Platforms
Abstract:
The recent switch to multi-core computing and the emergence of machine learning applications offer many opportunities for parallelization. A fundamental challenge is how to achieve scalability with the increasing number of cores. It is especially challenging for machine learning problems that have graph computational structure.
This thesis explores parallelism for exact inference in probabilistic graphical models to achieve scalable performance on multi-core platforms. Exact inference is widely used in probabilistic reasoning and machine learning. It also represents a large class of graph computations that have sophisticated computational patterns with large data sets. We propose parallel techniques to extract and exploit parallelism efficiently at multiple levels in the input graph.
• We first exploit parallelism available from the input graph using multithreading and task scheduling. At the node level, we explore data parallelism for the computational operations within a node. Data layout and data parallel algorithms are proposed for such node level computations. At the graph level, task parallelism is explored using directed acyclic graph (DAG) model. DAG scheduling is employed to efficiently map the tasks in the DAG to the hardware cores.
• In many cases, the input graph provides insufficient parallelism. To expose more parallelism, we study a relationship called 'weak dependency' between the tasks in a DAG. A novel DAG scheduling scheme is developed to exploit weak dependency for parallelism. In addition, pointer jumping technique is employed for exact inference when the input graph offers very limited parallelism due to its chain-like structure. With such explorations, a given fixed-size problem can still achieve high scalability with the increasing number of cores.
• In order to avoid the implementation complexity of many parallel techniques, we study the use of MapReduce as a high level programming model for exact inference. Our MapReduce-based algorithms for exact inference can also be applied for a class of graph computations with data dependency.
We implement and evaluate our techniques on state-of-the-art multi-core systems and demonstrate their scalability for a variety of input graphs.Location: Ronald Tutor Hall of Engineering (RTH) - 324
Audiences: Everyone Is Invited
Contact: Assistant to CS chair
-
CS Colloquium: Aditya Parameswaran (Stanford): Human-Powered Data Management
Tue, Apr 09, 2013 @ 03:30 PM - 05:00 PM
Thomas Lord Department of Computer Science
Conferences, Lectures, & Seminars
Speaker: Aditya Parameswaran, Stanford
Talk Title: Human-Powered Data Management
Series: CS Colloquium
Abstract: Fully automated algorithms are inadequate for a number of data analysis tasks, especially those involving images, video, or text. Thus, there is a need to combine "human computation" (or crowdsourcing), together with traditional computation, in order to improve the process of understanding and analyzing data. My thesis addresses several topics in the general area of human-powered data management. I design algorithms and systems for combining human and traditional computation for: (a) data processing, e.g., using humans to help sort, cluster, or clean data; (b) data extraction, e.g., having humans help create structured data from information in unstructured web pages; and (c) data gathering, i.e., asking humans to provide data that they know about or can locate, but that would be difficult to gather automatically. My focus in all of these areas is to find solutions that expend as few resources as possible (e.g., time waiting, human effort, or money spent), while still providing high quality results.
In this talk, I will first present a broad perspective of our research on human-powered data management, and I will describe some systems and applications that have motivated our research. I will then present details of one of the problems we have addressed: filtering large data sets with the aid of humans. Finally I will argue that human-powered data management is an area in its infancy, by describing a number of open problems I intend to address in my future research program.
Biography: Aditya Parameswaran is a Ph.D. student in the InfoLab at Stanford University, advised by Prof. Hector Garcia-Molina. He is broadly interested in data management, with research results in human computation, information extraction, and recommendation systems. Aditya is a recipient of the Key Scientific Challenges Award from Yahoo! Research (2010), two best-of-conference citations (VLDB 2010 and KDD 2012), and the Terry Groswith graduate fellowship at Stanford University.
Host: Cyrus Shahabi
Location: Seaver Science Library (SSL) - 150
Audiences: Everyone Is Invited
Contact: Assistant to CS chair
