Bounded coordinate-descent for biological sequence classification in high dimensional predictor space

TY - GEN

T1 - Bounded coordinate-descent for biological sequence classification in high dimensional predictor space

AU - Ifrim, Georgiana

AU - Wiuf, Carsten

PY - 2011/9/16

Y1 - 2011/9/16

N2 - We present a framework for discriminative sequence classification where linear classifiers work directly in the explicit high-dimensional predictor space of all subsequences in the training set (as opposed to kernel-induced spaces). This is made feasible by employing a gradient-bounded coordinatedescent algorithm for efficiently selecting discriminative subsequences without having to expand the whole space. Our framework can be applied to a wide range of loss functions, including binomial log-likelihood loss of logistic regression and squared hinge loss of support vector machines. When applied to protein remote homology detection and remote fold recognition, our framework achieves comparable performance to the state-of-the-art (e.g., kernel support vector machines). In contrast to state-of-the-art sequence classifiers, our models are simply lists of weighted discriminative subsequences and can thus be interpreted and related to the biological problem - a crucial requirement for the bioinformatics and medical communities.

AB - We present a framework for discriminative sequence classification where linear classifiers work directly in the explicit high-dimensional predictor space of all subsequences in the training set (as opposed to kernel-induced spaces). This is made feasible by employing a gradient-bounded coordinatedescent algorithm for efficiently selecting discriminative subsequences without having to expand the whole space. Our framework can be applied to a wide range of loss functions, including binomial log-likelihood loss of logistic regression and squared hinge loss of support vector machines. When applied to protein remote homology detection and remote fold recognition, our framework achieves comparable performance to the state-of-the-art (e.g., kernel support vector machines). In contrast to state-of-the-art sequence classifiers, our models are simply lists of weighted discriminative subsequences and can thus be interpreted and related to the biological problem - a crucial requirement for the bioinformatics and medical communities.

KW - Greedy coordinate-descent

KW - Logistic regression

KW - Sequence classification

KW - String classification

KW - Support vectormachines

UR - http://www.scopus.com/inward/record.url?scp=80052661040&partnerID=8YFLogxK

U2 - 10.1145/2020408.2020519

DO - 10.1145/2020408.2020519

M3 - Article in proceedings

AN - SCOPUS:80052661040

SN - 9781450308137

SP - 708

EP - 716

BT - Proceedings of the 17th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, KDD'11

T2 - 17th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, KDD'11

Y2 - 21 August 2011 through 24 August 2011

ER -