An Ensemble-Based Incremental Learning Approach to Data Fusion

Abstract

Abstract-This paper introduces Learn++, an ensemble of classifiers based algorithm originally developed for incremental learning, and now adapted for information/data fusion applications. Recognizing the conceptual similarity between incremental learning and data fusion, Learn++ follows an alternative approach to data fusion, i.e., sequentially generating an ensemble of classifiers that specifically seek the most discriminating information from each data set. It was observed that Learn++ based data fusion consistently outperforms a similarly configured ensemble classifier trained on any of the individual data sources across several applications. Furthermore, even if the classifiers trained on individual data sources are fine tuned for the given problem, Learn++ can still achieve a statistically significant improvement by combining them, if the additional data sets carry complementary information. The algorithm can also identify-albeit indirectly-those data sets that do not carry such additional information. Finally, it was shown that the algorithm can consecutively learn both the supplementary novel information coming from additional data of the same source, and the complementary information coming from new data sources without requiring access to any of the previously seen data.Index Terms-Data fusion, incremental learning, Learn++, multiple classifier/ensemble systems.

Functional Near-Infrared Spectroscopy and Electroencephalography: A Multimodal Imaging Approach

“…Behavioral performance was calculated as the percentage of presented stimuli that the subject correctly identified as targets or non-targets. Additionally, the two fusion approaches (and in particular the decision-level fusion, in agreement with [28]) provided an increase, albeit small, in the generalizing ability of the classifiers, as measured by a decrement in the accuracy standard deviation.…”

Data Fusion by Matrix Factorization

“…Data integration is an important aspect of bioinformatics and the subject of much recent research (Yu et al, 2010;Vaske et al, 2010;Mostafavi & Morris, 2012). It is expected that various data sources are complementary and that high accuracy of predictions can be achieved through data fusion (Parikh & Polikar, 2007;Pandey et al, 2010;Savage et al, 2010;Xing & Dunson, 2011). We studied the fusion of eleven di↵erent data sources to predict gene function in the social amoeba Dictyostelium discoideum and report on the cross-validated accuracy for 148 gene annotation terms (classes).…”

A New Divide and Conquer Based Classification for OCR

“…This is implemented with only a simple thresholding or comparison, whereas, the classification in new method is done on the total feature space by MLPs or KNNs. As Deviparikh's definition of classifier fusion in (Parikh et al, 2007), the new proposed method cannot be a classifier fusion, too. In fact, it is a new kind of classifier ensemble.…”