Hierarchical ensemble learning for multimedia categorization and autoannotation

Abstract

Abstract. This paper presents a hierarchical ensemble learning method applied in the context of multimedia autoannotation. In contrast to the standard multiple-category classification setting that assumes independent, non-overlapping and exhaustive set of categories, the proposed approach models explicitly the hierarchical relationships among target classes and estimates their relevance to a query as a trade-off between the goodness of fit to a given category description and its inherent uncertainty. The promising results of the empirical evaluation confirm the viability of the proposed approach, validated in comparison to several techniques of ensemble learning, as well as with different type of baseline classifiers.

Classification of Echolocation Calls from 14 Species of Bat by Support Vector Machines and Ensembles of Neural Networks

“…Classifiers in an ensemble can also be arranged into a hierarchy where the ensemble input is initially classified into a number of general classes before being classified to a more detailed class. This allows members of the ensemble to be trained on specific tasks, rather than the entire classification task [20][21]. Prudent selection of hierarchical classification branches can enhance overall classification performance.…”

Automatic Image Annotation with Relevance Feedback and Latent Semantic Analysis

“…The work in [21] and [22] builds and improves on a semantic ensemble which uses an ontology such as WordNet for determining a hierarchical representation of the semantic information for automatic annotation. One process specialises in propagating generic semantic information while the other process works to introduce specific knowledge about the document.…”

“…An ontology such as WordNet could be used to find parent concepts shared between labels to apply a more general annotation, as was done in [22]. For example, instead of the top-middle image in Figure 6 being labelled "bird tree insect lizard," when it is clearly a leopard, the common parent between "lizard," "bird," and "insect", which is "animal," could be used.…”

Automatic Annotation of Images from the Practitioner Perspective

“…A number of techniques have been reported which are designed to uncover the latent correlation between low-level visual features and high-level semantics [2,[8][9][10][11][12][13][14]. Typically such approaches involve a training set of pre-annotated images and the identification of visual features in the image such as blobs or salient objects.…”