Preference rules for label ranking: Mining patterns in multi-target relations

Sá, Cláudio Rebelo de; Azevedo, Paulo J.; Soares, Carlos; Jorge, Alípio Mário; Knobbe, Arno

doi:10.1016/j.inffus.2017.07.001

Cited by 8 publications

(3 citation statements)

References 37 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Specifically, given the label space Y, label ranking aims to assign each instance x with the correct ranking of all the labels, that is, a complete/partial, transitive, and asymmetric relation x defined on Y, where i x j means that the label i precedes the label j in the ranking associated with x. According to the taxonomy established by [26], label ranking methods can be divided into four categories: the ones decomposing the original problem to multiple simple objectives such as pointwise function [27], [28] and pairwise ranking loss [29], [30]; probabilistic methods including tree-based model [31], [32], [33], [34], [35], Gaussian mixture model [36] and structured learning [37]; the ones based on similarity [38], [39], [40]; and the rule-based ones [41], [42]. Please refer to the surveys for more details [26], [43].…”

Section: Label Rankingmentioning

confidence: 99%

Optimizing Partial Area Under the Top-K Curve: Theory and Practice

Wang

Yang

et al. 2022

IEEE Trans. Pattern Anal. Mach. Intell.

View full text Add to dashboard Cite

Top-k error has become a popular metric for large-scale classification benchmarks due to the inevitable semantic ambiguity among classes. Existing literature on top-k optimization generally focuses on the optimization method of the top-k objective, while ignoring the limitations of the metric itself. In this paper, we point out that the top-k objective lacks enough discrimination such that the induced predictions may give a totally irrelevant label a top rank. To fix this issue, we develop a novel metric named partial Area Under the top-k Curve (AUTKC). Theoretical analysis shows that AUTKC has a better discrimination ability, and its Bayes optimal score function could give a correct top-K ranking with respect to the conditional probability. This shows that AUTKC does not allow irrelevant labels to appear in the top list. Furthermore, we present an empirical surrogate risk minimization framework to optimize the proposed metric. Theoretically, we present (1) a sufficient condition for Fisher consistency of the Bayes optimal score function; (2) a generalization upper bound which is insensitive to the number of classes under a simple hyperparameter setting. Finally, the experimental results on four benchmark datasets validate the effectiveness of our proposed framework.

show abstract

Section: Label Rankingmentioning

confidence: 99%

Optimizing Partial Area Under the Top-K Curve: Theory and Practice

Wang

Yang

et al. 2022

IEEE Trans. Pattern Anal. Mach. Intell.

View full text Add to dashboard Cite

show abstract

“…Pattern mining: this stage is dedicated to finding a set of candidate patterns by an exploratory analysis using a search-space, which is defined by a set of inductive constraints provided by the user. There are several algorithms for mining patterns, those directly generating the patterns (e.g., rule miners) and those extracting patterns from the trees (e.g., from decision trees) [30], [31]. Pattern filtering: this stage focuses on selecting a subset of patterns coming from a large collection of patterns produced in the preceding stage.…”

Section: B Pattern-based Classificationmentioning

confidence: 99%

Cost-Sensitive Pattern-Based classification for Class Imbalance problems

et al. 2019

View full text Add to dashboard Cite

In several problems, contrast pattern-based classifiers produce high accuracy and provide an explanation of the result in terms of the patterns used for classification. However, class imbalance problems are a great challenge for these classifiers because there exist significantly fewer objects belonging to a class regarding the remaining classes and this biases the classification to the majority class. Therefore, in this paper, we propose an algorithm for discovering cost-sensitive patterns in class imbalance problems and a pattern-based classifier which uses these patterns for classification. Our proposal follows the idea of fusing pattern discovery with the cost-sensitive approach for class imbalance problems. Our experiments show that our proposal obtains cost-sensitive patterns, which allow attaining significantly lower misclassification cost than using patterns mined by other well-known state-of-the-art pattern miners. Also, we show that our proposed pattern-based classifier is suitable for working with cost-sensitive patterns. INDEX TERMSPattern-based classification, imbalanced databases, cost-sensitive problems. Data level: the methods following this approach create a balanced dataset from the imbalanced training dataset, either by reducing the majority class (Undersampling), by growing the minority class (Oversampling), or both (hybrid sampling). Those methods following this approach are known as resampling methods, and they impact the classifier's performance at the classification stage; either because of the inclusion of artificial objects that could produce overfitting or the exclusion of useful objects from the majority class [10]-[14]. Algorithm level: methods based on this approach modify a classifier trying to make it more accurate

show abstract

“…Several learning algorithms proposed for modeling Label Ranking data can be grouped as decomposition-based or direct (de Sá et al 2018). Decomposition methods divide the problem into several simpler problems (e.g., multiple binary problems).…”

Section: Label Rankingmentioning

confidence: 99%

Discovering a taste for the unusual: exceptional models for preference mining

et al. 2018

Self Cite

View full text Add to dashboard Cite

Exceptional preferences mining (EPM) is a crossover between two subfields of data mining: local pattern mining and preference learning. EPM can be seen as a local pattern mining task that finds subsets of observations where some preference relations between labels significantly deviate from the norm. It is a variant of subgroup discovery, with rankings of labels as the target concept. We employ several quality measures that highlight subgroups featuring exceptional preferences, where the focus of what constitutes 'exceptional' varies with the quality measure: two measures look for exceptional overall ranking behavior, one measure indicates whether a particular label stands out from the rest, and a fourth measure highlights subgroups with unusual pairwise label ranking behavior. We explore a few datasets and compare with existing techniques. The results confirm that the new task EPM can deliver interesting knowledge.

show abstract

Preference rules for label ranking: Mining patterns in multi-target relations

Cited by 8 publications

References 37 publications

Optimizing Partial Area Under the Top-K Curve: Theory and Practice

Optimizing Partial Area Under the Top-K Curve: Theory and Practice

Cost-Sensitive Pattern-Based classification for Class Imbalance problems

Discovering a taste for the unusual: exceptional models for preference mining

Contact Info

Product

Resources

About