The goal of multilabel (ML) classification is to induce models able to tag objects with the labels that better describe them. The main baseline for ML classification is Binary Relevance (BR), which is commonly criticized in the literature because of its label independence assumption. Despite this fact, this paper discusses some interesting properties of BR, mainly that it produces optimal models for several ML loss functions. Additionally, we present an analytical study about ML benchmarks datasets, pointing out some shortcomings. As a result, this paper proposes the use of synthetic datasets to better analyze the behavior of ML methods in domains with different characteristics. To support this claim, we perform some experiments using synthetic data proving the competitive performance of BR with respect to a more complex method in difficult problems with many labels, a conclusion which was not stated by previous studies.

Real-world applications demand effective methods to estimate the class distribution of a sample.In many domains, this is more productive than seeking individual predictions. At a first glance, the straightforward conclusion could be that this task, recently identified as quantification, is as simple as counting the predictions of a classifier. However, due to natural distribution changes occurring in real-world problems, this solution is unsatisfactory. Moreover, current quantification models based on classifiers present the drawback of being trained with loss functions aimed at classification rather than quantification. Other recent attempts to address this issue suffer certain limitations regarding reliability, measured in terms of classification abilities. This paper presents a learning method that optimizes an alternative metric that combines simultaneously quantification and classification performance. Our proposal offers a new framework that allows the construction of binary quantifiers that are able to accurately estimate the proportion of positives, based on models with reliable classification abilities.

No abstract

This paper presents a new approach for solving binary quantification problems based on nearest neighbor (NN) algorithms. Our main objective is to study the behavior of these methods in the context of prevalence estimation. We seek for NN-based quantifiers able to provide competitive performance while balancing simplicity and effectiveness. We propose two simple weighting strategies, PWK and PWK α , which stand out among state-of-the-art quantifiers. These proposed methods are the only ones that offer statistical differences with respect to less robust algorithms, like CC or AC. The second contribution of the paper is to introduce a new experiment methodology for quantification.

No abstract

Abstract-In many applications, the mistakes made by an automatic classifier are not equal, they have different costs. These problems may be solved using a cost-sensitive learning approach. The main idea is not to minimize the number of errors, but the total cost produced by such mistakes. This paper presents a new multiclass costsensitive algorithm, in which each example has attached its corresponding misclassification cost. Our proposal is theoretically well-founded and is designed to optimize costsensitive loss functions. This research was motivated by a real-world problem, the biomass estimation of several plankton taxonomic groups. In this particular application, our method improves the performance of traditional multiclass classification approaches that optimize the accuracy.

One approach to multi-class classification consists in decomposing the original problem into a collection of binary classification tasks. The outputs of these binary classifiers are combined to produce a single prediction. Winner-takesall, max-wins and tree voting schemes are the most popular methods for this purpose. However, tree schemes can deliver faster predictions because they need to evaluate less binary models. Despite previous conclusions reported in the literature, this paper shows that their performance depends on the organization of the tree scheme, i.e. the positions where each pairwise classifier is placed on the graph. Different metrics are studied for this purpose, proposing a new one that considers the precision and the complexity of each pairwise model, what makes the method to be classifier-dependent. The study is performed using Support Vector Machines (SVMs) as base classifiers, but it could be extended to other kind of binary classifiers. The proposed method, tested on benchmark data sets and on one real-world application, is able to improve the accuracy of other decomposition multi-class classifiers, producing even faster predictions.

No abstract