Implementing SVM in an RDBMS: Improved Scalability and Usability

Posted in Databases, Companies, Science on August 17, 2008


Implementing SVM in an RDBMS: Improved Scalability and Usability

 Lecture slides:

  • Overview
  • Data Mining in RDBMS
  • Database Infrastructure
  • SVM in the Database
  • SVM and the ODM Infrastructure
  • SVM Data Preparation Support
  • SVM Scalability Issues
  • Implemented Scalability
  • Additional Implemented
  • Working Set Selection
  • Who to Retain?
  • Who to Add?
  • Stratified Sampling
  • Small Model Generation
  • Build Scalability Results
  • Scoring Scalability Results
  • SVM Scoring as a SQL Operator
  • Ease of Use Issues
  • On-the-Fly SVM Parameter
  • Classification Accuracy
  • Regression Accuracy
  • Classification Capacity Estimate
  • Classification Standard Deviation
  • Regression Epsilon Estimate
  • Conclusions
  • Final Note

Author: Joseph S. Yarmus, Oracle

Watch Video

Tags: Databases, Oracle, Science, Lectures, Computer Science, Machine Learning, VideoLectures.Net, Kernel Methods