Boosting Self-Supervised Learning via Knowledge Transfer
dc.contributor.author | Noroozi, Mehdi | |
dc.contributor.author | Vinjimoor, Ananth | |
dc.contributor.author | Favaro, Paolo | |
dc.contributor.author | Pirsiavash, Hamed | |
dc.date.accessioned | 2019-07-03T14:28:50Z | |
dc.date.available | 2019-07-03T14:28:50Z | |
dc.date.issued | 2018-12-17 | |
dc.description | 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition | |
dc.description.abstract | In self-supervised learning, one trains a model to solve a so-called pretext task on a dataset without the need for human annotation. The main objective, however, is to transfer this model to a target domain and task. Currently, the most effective transfer strategy is fine-tuning, which restricts one to use the same model or parts thereof for both pretext and target tasks. In this paper, we present a novel framework for self-supervised learning that overcomes limitations in designing and comparing different tasks, models, and data domains. In particular, our framework decouples the structure of the self-supervised model from the final task-specific fine-tuned model. This allows us to: 1) quantitatively assess previously incompatible models including handcrafted features; 2) show that deeper neural network models can learn better representations from the same pretext task; 3) transfer knowledge learned with a deep model to a shallower one and thus boost its learning. We use this framework to design a novel self-supervised task, which achieves state-of-the-art performance on the common benchmarks in PASCAL VOC 2007, ILSVRC12 and Places by a significant margin. Our learned features shrink the mAP gap between models trained via self-supervised learning and supervised learning from 5.9% to 2.6% in object detection on PASCAL VOC 2007. | en_US |
dc.description.sponsorship | PF has been supported by the Swiss National Science Foundation (SNSF) grant number 200021 169622. HP has been supported by GE Research and Verisk Analytics. | en_US |
dc.description.uri | https://ieeexplore.ieee.org/document/8579073 | en_US |
dc.format.extent | 9 pages | en_US |
dc.genre | conference papers and proceedings preprints | en_US |
dc.identifier | doi:10.13016/m24lmq-2ssa | |
dc.identifier.citation | Mehdi Noroozi, et.al, Boosting Self-Supervised Learning via Knowledge Transfer, 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition, DOI: 10.1109/CVPR.2018.00975 | en_US |
dc.identifier.uri | https://doi.org/10.1109/CVPR.2018.00975 | |
dc.identifier.uri | http://hdl.handle.net/11603/14336 | |
dc.language.iso | en_US | en_US |
dc.publisher | IEEE | en_US |
dc.relation.isAvailableAt | The University of Maryland, Baltimore County (UMBC) | |
dc.relation.ispartof | UMBC Computer Science and Electrical Engineering Department Collection | |
dc.relation.ispartof | UMBC Faculty Collection | |
dc.rights | This item is likely protected under Title 17 of the U.S. Copyright Law. Unless on a Creative Commons license, for uses protected by Copyright Law, contact the copyright holder or the author. | |
dc.rights | © 2018 IEEE | |
dc.subject | learning (artificial intelligence) | en_US |
dc.subject | neural nets | en_US |
dc.subject | object detection | en_US |
dc.subject | knowledge transfer | en_US |
dc.subject | self-supervised learning | en_US |
dc.subject | incompatible models | en_US |
dc.subject | target domain | en_US |
dc.subject | effective transfer strategy | en_US |
dc.subject | deeper neural network models | en_US |
dc.subject | novel self-supervised task | en_US |
dc.title | Boosting Self-Supervised Learning via Knowledge Transfer | en_US |
dc.type | Text | en_US |