Learning SVMs from Sloppily Labeled Data

Stempfel, Guillaume; Ralaivola, Liva

doi:10.1007/978-3-642-04274-4_91

Cited by 45 publications

(30 citation statements)

References 3 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For instance, in [47] the hinge loss is replaced by a related loss function that takes into account the amount of noise in the data. With this loss function the optimization problem becomes non-convex.…”

Section: Robust Learning Algorithmsmentioning

confidence: 99%

See 1 more Smart Citation

Small margin ensembles can be robust to class-label noise

2015

View full text Add to dashboard Cite

Esta es la versión de autor del artículo publicado en: This is an author produced version of a paper published in:Neurocomputing 160 (2015) Subsampling is used to generate bagging ensembles that are accurate and robust to class-label noise. The effect of using smaller bootstrap samples to train the base learners is to make the ensemble more diverse. As a result, the classification margins tend to decrease. In spite of having small margins, these ensembles can be robust to class-label noise. The validity of these observations is illustrated in a wide range of synthetic and real-world classification tasks. In the problems investigated, subsampling significantly outperforms standard bagging for different amounts of class-label noise. By contrast, the effectiveness of subsampling in random forest is problem dependent. In these types of ensembles the best overall accuracy is obtained when the random trees are built on bootstrap samples of the same size as the original training data. Nevertheless, subsampling becomes more effective as the amount of class-label noise increases.

show abstract

Section: Robust Learning Algorithmsmentioning

confidence: 99%

“…For instance, the SVM's described [47,37] are tested in very specific cases: asymmetric noise [47] or data in which class probabilities are available [37]. An extensive empirical comparison of the different robust learning methods would be of great interest in the field.…”

Section: Robust Learning Algorithmsmentioning

confidence: 99%

Small margin ensembles can be robust to class-label noise

2015

View full text Add to dashboard Cite

show abstract

“…Finally, the comparison with StPMKL will give us insights into the effects of our simple modelling assumptions about the label flipping process in comparison with the model-free approach in StPMKL. We should mention that there exist label-noise robust versions of SVM [41,2] that could be a subject of further comparisons. However, these works again assume knowledge of label noise probability without providing an algorithm for automatically inferring the noise rate from data, and [41] limit themselves to linear kernels which means no procedure is proposed for the highly non-trivial problem of setting kernel parameters in label-noise conditions.…”

Section: Comparisons With State-of-the-art Classifiers and Other Labementioning

confidence: 99%

“…We should mention that there exist label-noise robust versions of SVM [41,2] that could be a subject of further comparisons. However, these works again assume knowledge of label noise probability without providing an algorithm for automatically inferring the noise rate from data, and [41] limit themselves to linear kernels which means no procedure is proposed for the highly non-trivial problem of setting kernel parameters in label-noise conditions. Although for these reasons a direct comparison with these methods would not be particularly meaningful, we believe that our approach for estimating these crucial parameters could be adapted to extend those methods as well in future developments.…”

Section: Comparisons With State-of-the-art Classifiers and Other Labementioning

confidence: 99%

Learning kernel logistic regression in the presence of class label noise

Bootkrajang

Kabán

2014

Pattern Recognition

View full text Add to dashboard Cite

The classical machinery of supervised learning machines relies on a correct set of training labels. Unfortunately, there is no guarantee that all of the labels are correct. Labelling errors are increasingly noticeable in today's classification tasks, as the scale and difficulty of these tasks increases so much that perfect label assignment becomes nearly impossible. Several algorithms have been proposed to alleviate the problem, of which a robust Kernel Fisher Discriminant is a successful example. However, for classification, discriminative models are of primary interest, and rather curiously, the very few existing label-robust discriminative classifiers are limited to linear problems.In this paper, we build on the widely used and successful kernelising technique to introduce a label-noise robust Kernel Logistic Regression classifier. The main difficulty that we need to bypass is how to determine the model complexity parameters when no trusted validation set is available. We propose to adapt the Multiple Kernel Learning approach for this new purpose, together with a Bayesian regularisation scheme. Empirical results on 13 benchmark data sets and two real-world applications demonstrate the success of our approach.

show abstract

“…Deep learning algorithms can be extended to handle label noise by additional network layers for noise modeling [45]. Robust kernels can be learned from the data in order to improve the effectiveness of kernel-based methods under label noise [7] and robust SVM methods have also been considered [44] [6].…”

Section: Related Workmentioning

confidence: 99%

Hubness-aware kNN classification of high-dimensional data in presence of label noise

Tomašev

Búza

2015

Neurocomputing

View full text Add to dashboard Cite

Learning with label noise is an important issue in classification, since it is not always possible to obtain reliable data labels. In this paper we explore and evaluate a new approach to learning with label noise in intrinsically high-dimensional data, based on using neighbor occurrence models for hubness-aware k-nearest neighbor classification. Hubness is an important aspect of the curse of dimensionality that has a negative effect on many types of similarity-based learning methods. As we will show, the emergence of hubs as centers of influence in high-dimensional data affects the learning process in presence of label noise. We evaluate the potential impact of hub-centered noise by defining a hubness-proportional random label noise model that is shown to induce a significantly higher kNN misclassification rate than the uniform random label noise. Real-world examples are discussed where hubness-correlated noise arises either naturally or as a consequence of an adversarial attack. Our experimental evaluation reveals that hubness-based fuzzy k-nearest neighbor classification and Naive Hubness-Bayesian k-nearest neighbor classification might be suitable for learning under label noise in intrinsically high-dimensional data, as they exhibit robustness to high levels of random label noise and hubness-proportional random label noise. The results demonstrate promising performance across several data domains.

show abstract

Learning SVMs from Sloppily Labeled Data

Cited by 45 publications

References 3 publications

Small margin ensembles can be robust to class-label noise

Small margin ensembles can be robust to class-label noise

Learning kernel logistic regression in the presence of class label noise

Hubness-aware kNN classification of high-dimensional data in presence of label noise

Contact Info

Product

Resources

About