*********************************
There is now a CONTENT FREEZE for Mercury while we switch to a new platform. It began on Friday, March 10 at 6pm and will end on Wednesday, March 15 at noon. No new content can be created during this time, but all material in the system as of the beginning of the freeze will be migrated to the new platform, including users and groups. Functionally the new site is identical to the old one. webteam@gatech.edu
*********************************
Istituto di Analisi dei Sistemi ed Informatica,
Consiglio Nazionale delle Ricerche, Roma, Italy
"Integer and Logic Programming for Knowledge Extraction and its application to Biological Data"
Abstract:
Knowledge Extraction and Data Mining techniques are used to identify relevant patterns and models in large amount of data. Several methods have been developed and successfully applied for a vast range of applications, and in this talk we focus on techniques that make use of Integer and Logic Programming to formulate and solve two central and strongly related problems in Knowledge Extraction: Feature Selection and Supervised Learning.
The method proposed for Feature Selection is based on integer programming formulations that represent the retained information using linear constraints associated with the observed data. A multistart iterative metaheuristic is proposed to efficiently solve the related optimization problems for instances of large size where standard optimization algorithms fail to obtain a solution of good quality.
The Supervised Learning method is based on the extraction of separating logic formulas from training data and converts the learning problem in a sequence of instances of a Minimum Cost Satisfiability problem (MINSAT), a well studied and hard optimization problem.
These two methods, when combined, are capable of extracting very synthetic information with a high semantic value. Their use appears to be particularly indicated for biological applications, where the experiments typically produce data sets of considerable size. In this talk we thus describe their applications to two problems that arise in the analysis of DNA and genomic sequences: Micorarray Analysis, for which we give results on a recent study on neurodegeneration of transgenic mice, and Barcode Analysis, a recent project supported by the Barcode of Life Consortium funded by the Sloan Foundation, that aims at classifying the species through a small fragment of mitochondrial DNA.
For more information please contact Dr. Alberto Apostolico.
