|
|
| Support Vector Machines (SVM) |
Since their introduction in 1992, Support Vector Machines marked the beginning of a new era in the learning from examples paradigm in artificial intelligence. Rooted in the Statistical Learning Theory developed by Vladimir Vapnik, Support Vector Machines quickly gained attention from the pattern recognition community due to a number of theoretical and computational merits.
Support Vector Machines represent a breakthrough in the theory of learning systems. Statistical Learning Theory, the backbone of Support Vector Machines, provides a new framework for modeling learning algorithms, merges the fields of machine learning and statistics, and inspires algorithms that overcome all of the above difficulties. A new generation of learning algorithms - or equivalently of statistical methods - has recently been developed, based on this theory.
Such methods prove remarkably resistant to the problems imposed by noisy data and high dimensionality. They are computationally efficient. The optimal solution can always be found. These methods have an inherent modular design that simplifies their implementation and analysis and allows the insertion of domain knowledge. More importantly, they come with theoretical guarantees about their generalization ability.
Recursive Feature Elimination - Support Vector Machine (RFE-SVM)
created by the scientific mathematicians currently at HDVY has been used by HDVY scientists to find discriminate relationships within clinical datasets as well as within gene expression datasets created from micro-arrays of tumor versus normal tissues. Using RFE-SVM HDVY scientists have been able to access information in micro-array datasets that the most advanced bioinformatics techniques missed. In one micro-array experiment RFE-SVM’S could filter irrelevant genes that were tissue specific rather than related to the malignancy. RFE-SVM has also been used to determine gene expression patterns that can be correlated to the severity of disease. The data analysis technique has been shown to improve diagnosis and prognosis by providing a physician with an enhanced decision tool. HDVY scientists feel that these analytic methods are effective for finding genes implicated in several cancers as well as assist with the pharmacogenetic and toxicological profiling of patients.
|
|
|
SVM PATENTS OWNED BY HDVY
Click on patent # to view entire patent
Patent # 6,128,608
Issued Oct. 3, 2000
Enhancing Knowledge Discovery using Multiple Support Vector Machines
Patent # 6,157,921
Issued Dec. 5, 2000
Enhancing Knowledge Discovery using Support Vector Machines in a Distributed Network Environment
Patent # 6,427,141
Issued Jul. 30, 2002
Enhancing Knowledge Discovery using Multiple Support Vector Machines
Patent # 6,658,395
Issued Dec. 2, 2003
Enhancing Knowledge Discovery from Multiple Data Sets using Multiple Support Vector Machines
Patent # 6,789,069
Issued Sep. 7, 2004
Method for enhancing knowledge discovered from biological data using a learning machine
Patent # 6,760,715
Issued Jul 6, 2004
Methods of identifying biological patterns using multiple support vector machines
Patent # 6,714,925
Issued Mar. 30, 2004
Methods of identifying biological patterns in a distributed network
Patent # 6,882,990
Issued Apr 26, 2005
Method for using SVM’s for processing multiple data sets
Application Ser. No.
10/056,438
Filed 1-23-2002
“Computer aided image analysis”
Application Ser. No.
10/057,849
Filed 1-24-2002
“Methods of Identifying Patterns in Biological Systems and Uses Thereof”
Application Ser. No.
10/087,145
Filed 3-1- 2002
“Spectral Kernels for Learning Machines”
Application No. PCT US02/14311
Filed 5-7-2002
“Kernels and Methods for Selecting Kernels for Use in Learning Machines”
Application No. PCT US02/16012
Filed 5-20-2002
“Methods for Feature Selection in a Learning Machine”
Application No. PCT US02/15666
Filed 5-17-2002
“Model Selection for Cluster Data Analysis”
Application No. PCT US02/19202
Filed 6-17-2002
“Data Mining Platform for Bioinformatics and Other Knowledge Discovery”
Application No. PCT US02/35576
Filed 11-7-2002
“Method for Feature Selection in a Support Vector Machine Using Feature Ranking”
RFE-SVM PATENTS
Patent # 6,789,069
Issued Sep. 7, 2004
Patent 10/057,849
Filed Jan 24, 2002
Click here to view complete intellectual property portfolio.
|
|
