Skip to Main content Skip to Navigation
Preprints, Working Papers, ...

Label Propagation for Deep Semi-supervised Learning

Ahmet Iscen 1 Giorgos Tolias 1 Yannis Avrithis 2 Ondrej Chum 1
1 VRG - Visual Recognition Group [Prague]
CTU/FEE - Faculty of electrical engineering [Prague]
2 LinkMedia - Creating and exploiting explicit links between multimedia fragments
Inria Rennes – Bretagne Atlantique , IRISA-D6 - MEDIA ET INTERACTIONS
Abstract : Semi-supervised learning is becoming increasingly important because it can combine data carefully labeled by humans with abundant unlabeled data to train deep neu-ral networks. Classic methods on semi-supervised learning that have focused on transductive learning have not been fully exploited in the inductive framework followed by modern deep learning. The same holds for the manifold assumption-that similar examples should get the same prediction. In this work, we employ a transductive label propagation method that is based on the manifold assumption to make predictions on the entire dataset and use these predictions to generate pseudo-labels for the unlabeled data and train a deep neural network. At the core of the transductive method lies a nearest neighbor graph of the dataset that we create based on the embeddings of the same network. Therefore our learning process iterates between these two steps. We improve performance on several datasets especially in the few labels regime and show that our work is complementary to current state of the art.
Complete list of metadatas

Cited literature [45 references]  Display  Hide  Download
Contributor : Yannis Avrithis <>
Submitted on : Tuesday, November 19, 2019 - 12:51:15 PM
Last modification on : Tuesday, February 25, 2020 - 8:08:12 AM
Document(s) archivé(s) le : Thursday, February 20, 2020 - 6:16:04 PM


Files produced by the author(s)


  • HAL Id : hal-02370207, version 1
  • ARXIV : 1904.04717


Ahmet Iscen, Giorgos Tolias, Yannis Avrithis, Ondrej Chum. Label Propagation for Deep Semi-supervised Learning. 2019. ⟨hal-02370207⟩



Record views


Files downloads