iCaRL: Incremental classifier and representation learning Conference Paper


Author(s): Rebuffi, Sylvestre-Alvise; Kolesnikov, Alexander; Sperl, Georg; Lampert, Christoph H
Title: iCaRL: Incremental classifier and representation learning
Affiliation IST Austria
Abstract: A major open problem on the road to artificial intelligence is the development of incrementally learning systems that learn about more and more concepts over time from a stream of data. In this work, we introduce a new training strategy, iCaRL, that allows learning in such a class-incremental way: only the training data for a small number of classes has to be present at the same time and new classes can be added progressively. iCaRL learns strong classifiers and a data representation simultaneously. This distinguishes it from earlier works that were fundamentally limited to fixed data representations and therefore incompatible with deep learning architectures. We show by experiments on CIFAR-100 and ImageNet ILSVRC 2012 data that iCaRL can learn many classes incrementally over a long period of time where other strategies quickly fail.
Conference Title: CVPR: Computer Vision and Pattern Recognition
Conference Dates: July 22 - July 25, 2017
Conference Location: Honolulu, HI, USA
Publisher: IEEE  
Date Published: 2017-04-14
Start Page: 5533
End Page: 5542
URL:
DOI: 10.1109/CVPR.2017.587
Notes: This work was in parts funded by the Eu- ropean Research Council under the European Union’s Seventh Framework Programme (FP7/2007-2013)/ERC grant agreement no 308036: ”Life-long learning of visual scene understanding” (L3ViSU). The Tesla K40 cards used for this research were do- nated by the NVIDIA Corporation
Open access: yes (repository)
IST Austria Authors
  1. Christoph Lampert
    87 Lampert
  2. Georg Sperl
    1 Sperl
Related IST Austria Work