Ángela Blanco

Combining dissimilarity based classifiers for cancer prediction using gene expression profiles

Presentacion oral, parte del suplemento: Highlights from the Third International Society for Computational Biology (ISCB) Student Council Symposium at the Fifteenth Annual International Conference on Intelligent Systems for Molecular Biology (ISMB) Vienna, Austria. 21 July 2007.

A local semi-supervised Sammon algorithm for textual data visualization

Sammon’s mapping is a powerful non-linear technique that allow us to visualize high dimensional object relationships. It has been applied to a broad range of practical problems and particularly to the visualization of the semantic relations among terms in textual databases. The word maps generated by the Sammon mapping suffer from a low discriminant power due to the well known “curse of dimensionality” and to the unsupervised nature of the algorithm. Fortunately the textual databases provide frequently a manually created classification for a subset of documents that may help to overcome this problem. In this paper we first introduce a modification of the Sammon mapping (SSammon) that enhances the local topology reducing the sensibility to the ’curse of dimensionality’. Next a semi-supervised version is proposed that takes advantage of the a priori categorization of a subset of documents to improve the discriminant power of the word maps generated. The new algorithm has been applied to the challenging problem of word map generation. The experimental results suggest that the new model improves significantly well known unsupervised alternatives.

Combining SVM classifiers for email anti-spam filtering

Spam, also known as Unsolicited Commercial Email (UCE) is becoming a nightmare for Internet users and providers. Machine learning techniques such as the Support VectorMachines (SVM) have achieved a high accuracy filtering the spam messages. However, a certain amount of legitimate emails are often classified as spam (false positive errors) although this kind of errors are prohibitively expensive. In this paper we address the problem of reducing particularly the false positive errors in anti-spam email filters based on the SVM. To this aim, an ensemble of SVMs that combines multiple dissimilarities is proposed. The experimental results suggest that the new method outperforms classifiers based solely on a single dissimilarity and a widely used combination strategy such as bagging.

Combining Dissimilarities in a Hyper Reproducing Kernel Hilbert Space for Complex Human Cancer Prediction

Support vector machines (SVM) have been applied to the classification of cancer samples using the gene expression profiles. However, they rely on Euclidean distances that fail to reflect accurately the proximities among sample profiles. Then, non Euclidean dissimilarities provide additional information that should be considered to reduce the misclassification errors. In this paper, we incorporate in the classical nu-SVM algorithm a linear combination of non-Euclidean dissimilarities. The weights of the combination are learnt in a HRKHS (hyper reproducing kernel Hilbert space) using an efficient semidefinite programming algorithm. This approach allow us to incorporate a smoothing term that penalizes the complexity of the family of distances and avoids overfitting. The experimental results suggest that the method proposed helps to reduce the misclassification errors in several human cancer problems.

Ensemble of Support Vector Machines to Improve the Cancer Class Prediction Based on the Gene Expression Profiles

DNA microarrays provide rich profiles that are used in cancer prediction considering the gene expression levels across a collection of samples.Support Vector Machines (SVM), have been applied to the classification of cancer samples with encouraging results. However, they are usually based on Euclidean distances that fail to reflect accurately the sample proximities. Besides, SVM classifiers based on non-Euclidean dissimilarities fail to reduce significantly the errors. In this paper, we propose an ensemble of SVM classifiers in order to reduce the errors. The diversity among classifiers is induced considering a set of complementary dissimilarities and kernels. The experimental results suggest that that our algorithm improves classifiers based on a single dissimilarity and a combination strategy such as Bagging.

Ensemble of dissimilarity based classifiers for cancerous samples classification

DNA Microarray technology allow us to identify cancerous tissues considering the gene expression levels across a collection of related samples. Several classifiers such as Support Vector Machines (SVM), k Nearest Neighbors (k-NN) or Diagonal Linear Discriminant Analysis (DLDA) have been applied to this problem. However, they are usually based on Euclidean distances that fail to reflect accurately the sample proximities. Several classifiers have been extended to work with non-Euclidean dissimilarities although none outperforms the others because they misclassify a different set of patterns. In this paper, we combine different kind of dissimilarity based classifiers to reduce the misclassification errors. The diversity among classifiers is induced considering a set of complementary dissimilarities for three different type of models. The experimental results suggest that the algorithm proposed helps to improve classifiers based on a single dissimilarity and a widely used combination strategy such as Bagging.

Classification of multiple cancer types in a Hyper Reproducing Kernel Hilbert Space

The classification of multiple cancer types based on the gene expression profiles is a challenging task. support vector machines (SVM) have been applied to this aim but they rely on Euclidean distances that fail to reflect accurately the proximities among sample profiles.

Boosting support vector machines using multiple dissimilarities

Support Vector Machines (SVM) are powerful machine learning techniques that are able to deal with high dimensional and noisy data. They have been successfully applied to a wide range of problems and particularly to the analysis of gene expression data. However SVM algorithms rely usually on the use of the Euclidean distance that often fails to reflect the object proximities. Several versions of the SVM have been proposed that incorporate non Euclidean dissimilarities. Nevertheless, different dissimilarities reflect complementary features of the data and no one can be considered superior to the others. In this paper, we present an ensemble of SVM classifiers that reduces the misclassification error combining different dissimilarities. The method proposed has been applied to identify cancerous tissues using Microarray gene expression data with remarkable results.

On the combination of dissimilarities for gene expression data analysis

DNA Microarray technology allows us to monitor the expression level of thousands of genes simultaneously. This technique has become a relevant tool to identify different types of cancer. Several machine learning techniques such as the Support Vector Machines (SVM) have been proposed to this aim. However, common SVM algorithms are based on Euclidean distances which do not reflect accurately the proximities among the sample profiles. The SVM has been extended to work with non-Euclidean dissimilarities. However, no dissimilarity can be considered superior to the others because each one reflects different features of the data. In this paper, we propose to combine several Support Vector Machines that are based on different dissimilarities to improve the performance of classifiers based on a single measure. The experimental results suggest that our method reduces the misclassification errors of classifiers based on a single dissimilarity and a widely used combination strategy such as Bagging.

Ensemble of Kernel Based Classifiers to Improve the Human Cancer Prediction using DNA Microarrays

DNA microarrays provide rich profiles that are used in cancer prediction considering the gene expression levels across a collection of samples. Support Vector Machines (SVM), have been applied to the classification of cancer samples with encouraging results. However, they are usually based on Euclidean distances that fail to reflect accurately the sample proximities. Besides, SVM classifiers based on non-Euclidean dissimilarities fail to reduce significantly the errors. In this paper, we propose an ensemble of SVM classifiers in order to reduce the misclassification errors. The diversity among classifiers is induced considering a set of complementary dissimilarities and kernels. The experimental results suggest that that our algorithm improves classifiers based on a single dissimilarity and a combination strategy such as Bagging.