The Lovász-Softmax loss: A tractable surrogate for the optimization of the intersection-over-union measure in neural networks

Autor: Maxim Berman, Amal Rannen Triki, Matthew B. Blaschko

Publikováno v: CVPR

The Jaccard index, also referred to as the intersection-over-union score, is commonly employed in the evaluation of image segmentation results given its perceptual qualities, scale invariance - which lends appropriate relevance to small objects, and

Externí odkaz: https://explore.openaire.eu/search/publication?articleId=doi_dedup___::7f56b06967e3392a9f726603244ae210
https://lirias.kuleuven.be/handle/123456789/621350

Joint embeddings of scene graphs and images

Autor: Belilovsky, E., Matthew Blaschko, Kiros, J. R., Urtasun, R., Zemel, R.

Publikováno v: Scopus-Elsevier
International Conference On Learning Representations-Workshop
International Conference On Learning Representations-Workshop, 2017, Toulon, France

International audience; Multimodal representations of text and images have become popular in recent years. Text however has inherent ambiguities when describing visual scenes, leading to the recent development of datasets with detailed graphical desc

Externí odkaz: https://explore.openaire.eu/search/publication?articleId=dedup_wf_001::3e36cbbe8e12131589a5ad3baa20f3e8
https://lirias.kuleuven.be/handle/123456789/581155

Testing for Differences in Gaussian Graphical Models: Applications to Brain Connectivity

Autor: Belilovsky, E., Varoquaux, G., Matthew Blaschko

Publikováno v: NIPS Proceedings
Neural Information Processing Systems (NIPS) 2016
Neural Information Processing Systems (NIPS) 2016, Dec 2016, Barcelona, Spain
Scopus-Elsevier

© 2016 NIPS Foundation - All Rights Reserved. Functional brain networks are well described and estimated from data with Gaussian Graphical Models (GGMs), e.g. using sparse inverse covariance estimators. Comparing functional connectivity of subjects

Externí odkaz: https://explore.openaire.eu/search/publication?articleId=doi_dedup___::8d78a03b11068859b06f40306f76b72a
https://inria.hal.science/hal-01248844

A Convex Surrogate Operator for General Non-Modular Loss Functions

Autor: Yu, J., Matthew Blaschko

Publikováno v: The 25th Belgian-Dutch Conference on Machine Learning
The 25th Belgian-Dutch Conference on Machine Learning, Sep 2016, Kortrijk, Belgium
The 19th International Conference on Artificial Intelligence and Statistics
The 19th International Conference on Artificial Intelligence and Statistics, May 2016, Cadiz, Spain
Scopus-Elsevier

Empirical risk minimization frequently employs convex surrogates to underlying discrete loss functions in order to achieve computational tractability during optimization. However, classical convex surrogates can only tightly bound modular loss functi

Externí odkaz: https://explore.openaire.eu/search/publication?articleId=doi_dedup___::5591d65d4e5040c06ed2393c57d45638
https://hal.inria.fr/hal-01417108

Efficient learning for discriminative segmentation with supermodular losses

Autor: Jiaqian Yu, Matthew B. Blaschko

Publikováno v: British Machine Vision Conference
British Machine Vision Conference, Sep 2016, York, United Kingdom
BMVC

International audience; Several supermodular losses have been shown to improve the perceptual quality of image segmentation in a discriminative framework such as a structured output support vector machine (SVM). These loss functions do not necessaril

Externí odkaz: https://explore.openaire.eu/search/publication?articleId=doi_dedup___::83a6759871463a487cb84252699cbdb2
https://lirias.kuleuven.be/handle/123456789/551162