December 2005
NSF study will probe the quintessence of surprise
Professor of computer science Pierre Baldi and information theorist Laurent Itti of the University of Southern California are joining with electro physiologist Douglas Muñoz of Queens University, Canada on a study of how brains measure “surprise” in a data stream.
Pierre
BaldiThe computational theory of surprise was developed by Baldi in 1999 and first published in 2002. The NSF award will allow the multi-organizational team to put the theory to test. Early laboratory experiments have shown the Bayesian model of surprise outperforms previous theories in predicting human responses to suddenly unusual stimuli.
“We believe this approach has potential for profoundly changing our current notions of how ‘important’ or ‘surprising’ information is understood,” said Baldi, also the Director of the Institute for Genomics and Informatics (IGB) at UCI.
Baldi and Itti will present the outlines of their new mathematical understanding of surprise — along with their empirical studies on human subjects that support their theory — on Wednesday, December 7 at the Neural Information Processing Systems (NIPS) Conference in Vancouver, British Columbia.
The new NSF project combines forces of Baldi’s computational and theory labs with the modeling and psychophysics lab of USC’s Laurent Itti, and expands the research to include neurophysiological studies of brain mechanisms in Muñoz electrophysiology lab at Queen’s University (Canada). Both experiments and modeling will be combined in the project to develop a new understanding of how the brain codes for novelty and importance.
November 2005
Welling receives grant to study learning taxonomies of the visual world
Assistant professor of computer science Max Welling along with Pietro Perona at Caltech have received a joint NSF grant to build a computer vision system that learns a taxonomy of visual object categories without supervision.
Max
WellingLearning visual object categories, and recognizing objects in images, is perhaps the most difficult and exciting problem in machine vision today. In light of the fast growing data deluge in science, engineering, industry and society, recognition systems must be able to operate without human supervision.
This poses new challenges: How can one learn automatically models of a large number of object classes from unlabelled images? How can one represent these object classes such that they can be searched efficiently? How can one leverage the learnt models to learn new object classes from very few examples?
Welling and Perona plan to combat these challenges by inferring hierarchical representations of object classes from unlabelled image data. Object classes are represented as constellations of parts, where each part extracts shape and appearance information.
Non-parametric Bayesian techniques may be employed to organize these object classes into tree-structured representations. The richness of this representation grows incrementally as more data is presented to the system. New similarity measures between object classes naturally derive from this representation facilitating recognition.
October 2005
Professors awarded grant for satellite image analysis
Professor of computer science Padhraic Smyth along with professor of earth systems science Gudrun Magnusdottir and professor of statistics Hal Stern have been awarded a 3-year $618,000 National Science Foundation grant to develop statistical image analysis and tracking algorithms for analyzing satellite images of large-scale atmospheric disturbances over time.
Padhraic
SmythThe project will focus specifically on developing tools to better understand a phenomenon known as the Inter-Tropical Convergence Zone (ITCZ), an important part of global large-scale atmospheric circulation. New models and algorithms will be investigated for tracking dynamic objects in images over time, using historical image data from visible and infra-red imagery, as well as satellite estimates of winds and liquid cloud water content.
Research on this project will be carried out by a collaborative effort between the Bren School of ICS and the Department of Earth System Sciences at UC Irvine, including faculty members Smyth and Stern in ICS and Magnusdottir (in Earth System Sciences), Ph.D. student Lucas Scharenbroich and postdoctoral fellow Alex Ihler in the Computer Science department, and postoctoral fellow Chai-chi Wang in the ESS department.
September 2005
Suda receives funding to pioneer the new field of molecular communication
Tatsuya Suda, professor of computer science, has received $100,000 in funding from the National Science Foundation and $741,000 from the National Institute of Information and Communications Technology in Japan to develop the new research area of molecular communication by designing communication networks between biological nano-machines over the 2005-2006 academic year.
Tatsuya
SudaMolecular communication is a new and emerging area that integrates computer science, biological science and nano-science. The goal is to apply communication and signaling mechanisms found in biological entities (such as cells) to communication between nano-scale biological machines.
Applying mechanisms found in biological entities is the most promising approach, as it is extremely difficult to create nano-scale electrical and optical devices and power sources, thus virtually eliminating the possibility of extending the current communication technology to nano-scale communication.
Molecular communication requires an understanding of biological signaling and communication mechanisms in order to artificially recreate such mechanisms for communication between nano-scale machines that are made of bio-materials and non-bio/semi-bio-friendly materials.
Suda will work on establishing a new theoretical foundation for nano-scale communication, including the development of new information theory to measure the information carried in nano-scale communication and of new coding theory to map the information onto carrier signals used in biological entities such as protein and CA ions.
This project will also require new simulation techniques for nano-scale communication and the design of components and devices such as a sender, encoder, receiver, decoder, and various logical gates.
August 2005
Professors awarded $1.8 million NIH grant
Rick
LathropThe National Cancer Institute (NCI) of the National Institutes of Health (NIH) has awarded three Institute for Genomics and Bioinformatics (IGB) researchers $1.8 million to pursue cancer research.
The interdisciplinary team includes Richard Lathrop and Pierre Baldi of the Department of Computer Science and Rainer Brachmann of the Division of Hematology/Oncology, Department of Medicine. The IGB-seeded project applies machine learning to the central tumor-suppressor protein known as “p53 .” The ultimate goal of the project is to identify new anti-cancer drugs.
Lathrop, Baldi and Brachmann represent a computationally focused team within a large collaborative group of UCI researchers studying p53, p53 mutants and their functional rescue. Fostered by Dr. Brachmann, the interdisciplinary effort includes four schools and nine laboratories. Brachmann observed that making changes to specific amino acids of a damaged, cancer-causing p53 protein often restores normal function.
The new grant, part of the NIH “Innovations in Biomedical Computational Science and Technology” (BISTI) program, will allow the computational team to pursue machine learning and experimental investigations on p53 activities.
The research team expects to be better able to predict how function can be restored to damaged p53 protein, found in close to half of all human cancers. This information will be key to the development of anti-cancer drugs to treat human cancers with p53 mutations, known to be particularly resistant to current treatment options.
Dechter awarded Radcliffe fellowship
Rina Dechter, professor of computer science, has been awarded a fellowship from the Radcliffe Institute for Advanced Study at Harvard University to pursue her research in automated reasoning.
Rina
DechterDechter will spend the 2005-2006 academic year at the Radcliffe Institute to pursue her project on advanced search strategies for mixed probabilistic and deterministic networks.
Dechter's research has focused on automated reasoning, i.e., writing computer programs that assist in solving problems and answering questions by drawing conclusions from knowledge (e.g., data and facts). Automated reasoning research has evolved around two distinct ways of representing knowledge. One way of representing knowledge uses hard constraints.
An example of a hard constraint might be "a student cannot enroll in two courses if they meet at the same time." The other recognizes that knowledge involves uncertainty and therefore uses probabilistic statements such as "if a student has a solid math background, he or she is likely
(but not certain) to do well in computer classes."Many real-life applications involve both types of information. The goal of Dechter's research project during her fellowship year is to work on a framework for automated reasoning that would accommodate both hard constraints and probabilistic information.
Reasoning about such "mixed" knowledge bases often requires lengthy and complex computation. Dechter will therefore focus on the development of efficient computational techniques that will enhance the usefulness of such automated reasoning systems.
Radcliffe Institute fellowships are designed to support scholars, scientists, artists, and writers of exceptional promise and demonstrated accomplishments who wish to pursue work in academic and professional fields and in the creative arts.
To learn more about Dechter's work, visit her website
at www.ics.uci.edu/~dechter.Also visit the the Radcliffe Institute website for additional information about the program and Dechter's work.
Paper wins award at Eurographics conference
"Hierarchyless Simplification, Stripification, and Compression of Triangulated Two-Manifolds" authored by the second year graduate student, Pablo Diaz-Gutierrez and computer science professors Gopi Meenakshisundaram and Renato Pajarola won the second best paper award at the annual Eurographics conference held in Dublin, Ireland August 26th - September 1st, 2005.
With more than 300 papers submitted, the acceptance rate for this conference was around 15 percent. The award is based on three criteria, the importance of results, clarity of the written paper, and the clarity of the final presentation. In the same conference last year, held in Grenoble, France, a paper by computer sceince professors Gopi Meenakshisundaram and David Eppstein also won the second best paper award.
May 2005
Welling receives NSF CAREER award
Max Welling, assistant professor of computer science, received a $450,000 NSF CAREER award. The award will fund research into a new class of probabilistic "undirected bipartite graphical models'' (UBG), that extract meaningful representations of input data.
Max
WellingThe special structure of UBG models allows handling of large data-sets and very fast query processing which could make them useful in image restoration and information retrieval.
NSF's Faculty Early Career Development (CAREER) award, recognizes a young researcher's dual commitment to scholarship and education.
June 2005
Dutt appointed Editor-in-Chief of ACM-TODAES
Nikil
DuttNikil Dutt, professor of computer science, has been appointed Editor in-Chief of ACM Transactions on Design Automation of Electronic Systems (ACM-TODAES).
ACM-TODAES is the premier journal in the area of design automation and provides a comprehensive coverage of innovative works concerning the specification, design, analysis, simulation, testing and evaluation of very large-scale integrated electronic systems, emphasizing a computer science/engineering orientation.
The publication can be viewed online at: http://www.acm.org/todaes/.
UC copyright | UCI directory | Intranet | Site Map | icswebmaster@ics.uci.edu | Last update: November 01 2006