New Fast Algorithms for Variable Selectionbased on Classi
نویسندگان
چکیده
Variable selection is an important methodology in multivariate statistics, especially in the context of classiication. However, because the direct evaluation of the subsets using a classiier has been computationally too expensive in the past for a medium to large number of variables, variable selection has instead been performed using simple measures of class separation such as Wilk's or the Mahalanobis distance. We present new fast algorithms for quadratic and linear classiiers with time complexities which, to within a constant, are the same as those for the above mentioned heuristics. Comparing the new algorithms to previous implementations, we show that dramatic speed-ups are achieved.
منابع مشابه
Extending Fast Nearest Neighbour Search Algorithms for Approximate k-NN Classification
The nearest neighbour (NN) and k-nearest neighbour (kNN) classi cation rules have been widely used in pattern recognition due to its simplicity and good behaviour. Exhaustive nearest neighbour search can become unpractical when facing large training sets, high dimensional data or expensive similarity measures. In the last years a lot of NN search algorithms have been developed to overcome those...
متن کاملDevelopment of the Tournament Selection-based Genetic Algorithms with the Aid of a New Selection Method
This paper proposes a method that combines tournament selectionbased genetic algorithms (GA) with simulated annealing (SA) to improve the performance of GA. Furthermore random signal-based learning (RSBL) is added as an additional GA operator to refine the solutions after mutation. The fuzzy controller for the inverted pendulum is used to verify the validity of the proposed method.
متن کاملImproving Service Utility and Service Delivery in Next Generation Networks
Analysis of Joint Call Admission Control Strategies for Heterogeneous Cellular Networks Olabisi Falowo (University of Cape Town, South Africa); H. Anthony Chan (University of Cape Town, South Africa); “This paper investigates three joint call admission control algorithms (JCAC) for heterogeneous cellular networks. These algorithms are random-selectionbased, service-based, and load-based JCAC al...
متن کاملgpALIGNER: A Fast Algorithm for Global Pairwise Alignment of DNA Sequences
Bioinformatics, through the sequencing of the full genomes for many species, is increasingly relying on efficient global alignment tools exhibiting both high sensitivity and specificity. Many computational algorithms have been applied for solving the sequence alignment problem. Dynamic programming, statistical methods, approximation and heuristic algorithms are the most common methods appli...
متن کاملSPRINT: A Scalable Parallel Classi er for Data Mining
Classi cation is an important data mining problem. Although classi cation is a wellstudied problem, most of the current classication algorithms require that all or a portion of the the entire dataset remain permanently in memory. This limits their suitability for mining over large databases. We present a new decision-tree-based classi cation algorithm, called SPRINT that removes all of the memo...
متن کامل