Two Stage Classifier Chain Architecture for efficient pair-wise multi-label learning

Gjorgjevikj, Dejan; Madjarov, Gjorgji

doi:10.1109/mlsp.2011.6064599

Cited by 5 publications

(3 citation statements)

References 15 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We compare the proposed ABC-based stacking algorithm with five state-of-the-art ensemble multilabel classification algorithms on all 10 benchmark datasets. The ten base level classifiers are CC [6], ECC [7], MLkNN [8], RAkEL [9], EPS [10], BRkNN [31], TSVA [32], TSCCA [33], IBLR [34], and CLR [35]. The five ensemble multilabel classification algorithms are RAkEL [9], ECC [6,7], EPS [10], RF-PCT [12], and ML-Forest [11].…”

Section: Methodsmentioning

confidence: 99%

ABC-based stacking method for multilabel classification

Ding¹,

Wu²

2019

Turk J Elec Eng & Comp Sci

View full text Add to dashboard Cite

Multilabel classification is a supervised learning problem wherein each individual instance is associated with multiple labels. Ensemble methods are effective in managing multilabel classification problems by creating a set of accurate, diverse classifiers and then combining their outputs to produce classifications. This paper presents a novel stacking-based ensemble algorithm, ABC-based stacking, for multilabel classification. The artificial bee colony algorithm, along with a single-layer artificial neural network, is used to find suitable meta-level classifier configurations.The optimization goal of the meta-level classifier is to maximize the average accuracy of classification of all the instances involved. We run an experiment on 10 benchmark datasets of varying domains and compare the proposed approach to five other ensemble algorithms to demonstrate the feasibility and effectiveness of ABC-based stacking.

show abstract

Section: Methodsmentioning

confidence: 99%

ABC-based stacking method for multilabel classification

Ding¹,

Wu²

2019

Turk J Elec Eng & Comp Sci

View full text Add to dashboard Cite

show abstract

“…We evaluate our structured prediction learning framework in multi-label classification settings, as described in Section 5.2. We use the yeast (Elisseeff and Weston, 2001) dataset, which has 1500 training and 917 test instances with d = 103 attributes and |C|= 103 classes, and the scene (Gjorgjevikj and Madjarov, 2011) dataset, which has 1211 train and 1196 test instances with d = 294 attributes and |C|= 6 classes. For each input x, we construct the edge features R(x) (Equation (20)) by a dimensionality reduction of x.…”

Section: Multi-label Classification Datasetsmentioning

confidence: 99%

Discriminative Training of Conditional Random Fields with Probably Submodular Constraints

Berman

Blaschko

2020

Int J Comput Vis

View full text Add to dashboard Cite

Problems of segmentation, denoising, registration and 3D reconstruction are often addressed with the graph cut algorithm. However, solving an unconstrained graph cut problem is NP-hard. For tractable optimization, pairwise potentials have to fulfill the submodularity inequality. In our learning paradigm, pairwise potentials are created as the dot product of a learned vector w with positive feature vectors. In order to constrain such a model to remain tractable, previous approaches have enforced the weight vector to be positive for pairwise potentials in which the labels differ, and set pairwise potentials to zero in the case that the label remains the same. Such constraints are sufficient to guarantee that the resulting pairwise potentials satisfy the submodularity inequality. However, we show that such an approach unnecessarily restricts the capacity of the learned models. Guaranteeing submodularity for all possible inputs, no matter how improbable, reduces inference error to effectively zero, but increases model error. In contrast, we relax the requirement of guaranteed submodularity to solutions that are probably approximately submodular. We show that the conceptually simple strategy of enforcing submodularity on the training examples guarantees with low sample complexity that test images will also yield submodular pairwise potentials. Results are presented in the binary and muticlass settings, showing substantial improvement from the resulting increased model capacity.

show abstract

“…Typically, they rely on extending the attribute vectors by information about the classes to which the given example has already been shown to belong [3], [10], [12], [13], [14], [15], [16]. In our opinion, though, most of these papers fail to pay adequate attention to two critical issues: error-propagation, and unnecessary relationships.…”

Section: Introductionmentioning

confidence: 99%

PruDent: A Pruned and Confident Stacking Approach for Multi-Label Classification

Al‐Ali

Kubát

2015

IEEE Trans. Knowl. Data Eng.

View full text Add to dashboard Cite

Over the past decade or so, several research groups have addressed the problem of multi-label classification where each example can belong to more than one class at the same time. A common approach, called Binary Relevance (BR), addresses this problem by inducing a separate classifier for each class. Research has shown that this framework can be improved if mutual class dependence is exploited: an example that belongs to class X is likely to belong also to class Y ; conversely, belonging to X can make an example less likely to belong to Z. Several works sought to model this information by using the vector of class labels as additional example attributes. To fill the unknown values of these attributes during prediction, existing methods resort to using outputs of other classifiers, and this makes them prone to errors. This is where our paper wants to contribute. We identified two potential ways to prune unnecessary dependencies and to reduce error-propagation in our new classifier-stacking technique, which is named PruDent.Experimental results indicate that the classification performance of PruDent compares favorably with that of other state-of-the-art approaches over a broad range of testbeds. Moreover, its computational costs grow only linearly in the number of classes.

show abstract

Two Stage Classifier Chain Architecture for efficient pair-wise multi-label learning

Cited by 5 publications

References 15 publications

ABC-based stacking method for multilabel classification

ABC-based stacking method for multilabel classification

Discriminative Training of Conditional Random Fields with Probably Submodular Constraints

PruDent: A Pruned and Confident Stacking Approach for Multi-Label Classification

Contact Info

Product

Resources

About