Obtaining scalable and accurate classification in large-scale spatio-temporal domains

Abstract

We present an approach for learning models that obtain accurate classification of data objects, collected in large-scale spatio-temporal domains. The model generation is structured in three phases: spatial dimension reduction, spatio-temporal features extraction, and feature selection. Novel techniques for the first two phases are presented, with two alternatives for the middle phase. We explore model generation based on the combinations of techniques from each phase. We apply the introduced methodology to data-sets from the Voltage-Sensitive Dye Imaging (VSDI) domain, where the resulting classification models successfully decode neuronal population responses in the visual cortex of behaving animals. VSDI is currently the best technique enabling simultaneous high spatial (10,000 points) and temporal (10 ms or less) resolution imaging from neuronal population in the cortex. We demonstrate that not only our approach is scalable enough to handle computationally challenging data, but it also contributes to the neuroimaging field of study with its decoding abilities. The effectiveness of our methodology is further explored on a data-set from the hurricanes domain, and a promising direction, based on the preliminary results of hurricane severity classification, is revealed.

Foundations of Feature Selection

“…For this reason, a typical approach is to find a method that works satisfactorily for a specific problem setting. New feature selection methods are constantly emerging using different strategies: a) combining several feature selection methods, which could be done by using algorithms from the same approach, such as two filters [28], or coordinating algorithms from two different approaches, usually filters and wrappers [29,30]; b) combining feature selection approaches with other techniques, such as feature extraction [31] or tree ensembles [32]; c) reinterpreting existing algorithms [33], sometimes to adapt them to specific problems [34]; d) creating new methods to deal with still unresolved situations [35,36]; and e) using an ensemble of feature selection techniques to ensure better behavior [37,38].…”

Feature Selection Algorithms for Malaysian Dengue Outbreak Detection Model

“…Past research has shown that subset of features could represent the data more efficiently and leads to better understanding (Vainer et al 2011). Datasets with irrelevant and redundant information could lead to a complicated learning process and less accurate results (Tuv et al 2009).…”

“…Zhang et al (2008) apply reliefF and mRMR as a combination of different algorithms from the same approach (filters), whereas El Akadi et al (2011) apply mRMR filter and GA wrapper using filters and wrappers approaches. On the other hand, other works combined feature selection with other processes, such as combining spatial dimension reduction, spatiotemporal features extraction, and feature selection for hurricane severity classification (Vainer et al 2011). Tuv et al (2009 proposed feature selection with ensembles, artificial variables and redundancy elimination.…”

Enhanced clustering analysis pipeline for performance analysis of parallel applications

“…However, most researchers agree that there is not a so-called "best method" and their efforts are focused on finding a good method for a specific problem set. Therefore, the recent feature selection methods are constantly appearing using different strategies: a) combining several feature selection methods, which could be done by using algorithms from the same approach, such as two filters [130], or coordinating algorithms from two different approaches, usually, filters and wrappers [131,132] b) combining feature selection approaches with other techniques, such as feature extraction [133] or tree ensembles [134]; c) reinterpreting existing algorithms [135], sometimes to adapt them to specific problems [136]; d) creating methods to deal with still unresolved situations [137,138] using an ensemble of feature selection techniques to ensure a better behavior [139,140]).…”