Minimisation du risque empirique avec des fonctions de perte nonmodulaires

Doctoral thesis French OPEN
Yu, Jiaqian;
(2017)
  • Publisher: HAL CCSD
  • Subject: Structured Prediction | Loss Function | [ SPI.OTHER ] Engineering Sciences [physics]/Other | La fonction de perte de substitution | Surrogate Loss Function | Fonction de perte | [SPI.OTHER]Engineering Sciences [physics]/Other | Submodular | Prédiction structurée | Supermodular

This thesis addresses the problem of learning with non-modular losses. In a prediction problem where multiple outputs are predicted simultaneously, viewing the outcome as a joint set prediction is essential so as to better incorporate real-world circumstances. In empiri... View more
  • References (1)

    6.1 Contributions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101 6.2 Publication list . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104 1.1 Examples of structured prediction problem in nature language processing and images segmentation problem. . . . . . . . . . . . . . . . . . . . . . . . 1.2 Examples of training samples for an image segmentation problem where the foreground objects are people. Images are taken from the Pascal VOC dataset [Everingham et al., 2010] . . . . . . . . . . . . . . . . . . . . . . . . 1.3 Set functions in general. The intersection of submodular functions and supermodular functions, is the set of modular functions. In general, there are many functions that are neither submodular nor supermodular. . . . . . . . 1.4 The structure of this thesis in a Venn diagram related to the submodularity of the loss functions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1 An example on multi-label prediction problem as a set prediction problem.

  • Related Research Results (1)
    Inferred by OpenAIRE
    software
    aistats2016 software on GitHub
    72%
  • Metrics
Share - Bookmark