Model selection for cluster data analysis

a cluster data and model selection technology, applied in the field of learning machines, can solve the problems of increasing difficulty in facilitating human comprehension of the information in this data, unapproachable problems, and accelerating growth in the system and method of generating, collecting and storing vast amounts of data, and achieve the effect of facilitating the clustering task and low noise level

Inactive Publication Date: 2005-03-31
HEALTH DISCOVERY CORP +1
View PDF0 Cites 38 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Benefits of technology

In one embodiment, the inventive method provides means for extraction of information from gene expression profiles, in particular, extraction of the most relevant clusters of temporal expression profiles from tens of thousands of measured profiles. In one variation of the present embodiment, the clustering algorithm is the k-means algorithm. Other embodiments include all pairwise clustering methods (e.g., hierarchical clustering) and support vector clustering (i.e., a support vector machine (SVM)). In one variation, the fit involves affine transformations such as translation, rotation, and scale. Other transformations could b

Problems solved by technology

Recent advancements in database technology have lead to an explosive growth in systems and methods for generating, collecting and storing vast amounts of data.
While database technology enables efficient collection and storage of large data sets, the challenge of facilitating human comprehension of the information in this data is growing ever more difficult With many existing techniques the problem has become unapproachable.
Currently, there are no methods, systems or devices for adequately analyzing the data generated by such biological investigations into the genome and proteome.
Clustering analysis is unsupervised learning—it is done without suggestion from an external supervisor; classes and training examples are not given a priori.
However, most c

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Model selection for cluster data analysis
  • Model selection for cluster data analysis
  • Model selection for cluster data analysis

Examples

Experimental program
Comparison scheme
Effect test

example 1

Gaussian Data

Referring first to FIGS. 1a-1c, FIG. 1a shows a mixture of four Gaussians. The histograms of the score for varying values of k for this data is plotted in FIG. 1b. Histograms are shown for each value of k in the range of 2 to 7. Observations regarding the histograms are that at k=2, there is a peak at 1, since almost all the runs discriminated between the two upper and two lower clusters. At k=3, most runs separated the two lower clusters, and at k=4 most runs found the “correct” clustering as is reflected in the distribution of scores that is still close to 1.0. At k>4 there is no longer essentially one preferred solution. There is, in fact, a wide variety of solutions, evidenced by the widening spectrum of the similarities. FIG. 1c plots the cumulative distributions of the correlation score for each k, where k=2 at the rightmost side of the plot (at peak 1), and k=7 being the leftmost curve.

example 2

DNA Microarray Data

The next dataset considered was the yeast DNA microarray data of M. Eisen et al. (“Genetics cluster analysis and display of genome-wide expression patterns”, Proc. Natl. Acad. Sci. USA, 95: 14863-14868, December 1998. The data is a matrix which represents the mRNA expression levels of n genes across a number of experiments. Some of the genes in the data have known labels according to a functional class. Five functional classes were selected along with genes that belong uniquely to these five functional classes. This yielded a dataset with 208 genes, with 79 features (experiments). Data was normalized by subtracting the mean and dividing by the standard deviation for each column. This was also performed for the rows, and repeated for the columns. At this stage the first three principal components were extracted. The distribution and histogram of scores is give in FIG. 2a fork over the range of 2 to 7. The same behavior is observed as seen in the mixture of four G...

example 3

Uniformly Distributed Data

The results of a test run on data uniformly distributed on the unit cube is shown in FIGS. 3a and 3b. The distributions are quite similar to each other, with no change that can be interpreted as a transformation from a stable set of solutions to unstable solutions.

The preceding examples indicate a simple way for choosing k as the value where there is a transition from a score distribution that is concentrated near 1 to a wider distribution. This can be quantified, e.g., by an increase in the area under the cumulative distribution function or by an increase in

S(K)=P(s>0.90).

The value of 0.9 is arbitrary, but any value close to 1 would work on the set of examples considered here.

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

PUM

No PUM Login to view more

Abstract

A model selection method is provided for choosing the number of clusters, or more generally the parameters of a clustering algorithm. The algorithm is based on comparing the similarity between pairs of clustering runs on sub-samples or other perturbations of the data. High pairwise similarities show that the clustering represents a stable pattern in the data. The method is applicable to any clustering algorithm, and can also detect lack of structure. We show results on artificial and real data using a hierarchical clustering algorithm.

Description

FIELD OF THE INVENTION The present invention relates to the use of learning machines to identify relevant patterns in datasets containing large quantities of diverse data, and more particularly to a method and system for unsupervised learning for determining an optimal number of data clusters into which data can be divided to best enable identification of relevant patterns. BACKGROUND OF THE INVENTION Knowledge discovery is the most desirable end product of data collection. Recent advancements in database technology have lead to an explosive growth in systems and methods for generating, collecting and storing vast amounts of data. While database technology enables efficient collection and storage of large data sets, the challenge of facilitating human comprehension of the information in this data is growing ever more difficult With many existing techniques the problem has become unapproachable. Thus, there remains a need for a new generation of automated knowledge discovery tools....

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

Application Information

Patent Timeline
no application Login to view more
IPC IPC(8): G06G7/48G06G7/58G06K9/62G16B25/10G16B40/20G16B40/30
CPCG06F19/20G06F17/30705G06K9/6218G06F19/24G06F16/35G16B25/00G16B40/00G16B40/30G16B40/20G16B25/10G06F18/23
Inventor BEN-HUR, ASAELISSEEFF, ANDREGUYON, ISABELLE
Owner HEALTH DISCOVERY CORP
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap