@inproceedings{conf/mir/SpringsteinE16, added-at = {2018-11-14T00:00:00.000+0100}, author = {Springstein, Matthias and Ewerth, Ralph}, biburl = {https://www.bibsonomy.org/bibtex/205e84ffa33e4a3bfcc8b5bd4aa244cd9/dblp}, booktitle = {ICMR}, crossref = {conf/mir/2016}, editor = {Kender, John R. and Smith, John R. and Luo, Jiebo and Boll, Susanne and Hsu, Winston H.}, ee = {https://www.wikidata.org/entity/Q57757303}, interhash = {adaf35cfb79014cc03925a5561cb5a5c}, intrahash = {05e84ffa33e4a3bfcc8b5bd4aa244cd9}, isbn = {978-1-4503-4359-6}, keywords = {dblp}, pages = {377-380}, publisher = {ACM}, timestamp = {2018-11-15T15:47:08.000+0100}, title = {On the Effects of Spam Filtering and Incremental Learning for Web-Supervised Visual Concept Classification.}, url = {http://dblp.uni-trier.de/db/conf/mir/icmr2016.html#SpringsteinE16}, year = 2016 } @inproceedings{Springstein:2016:ESF:2911996.2912072, abstract = {Deep neural networks have been successfully applied to the task of visual concept classification. However, they require a large number of training examples for learning. Although pre-trained deep neural networks are available for some domains, they usually have to be fine-tuned for an envisaged target domain. Recently, some approaches have been suggested that are aimed at incrementally (or even endlessly) learning visual concepts based on Web data. Since tags of Web images are often noisy, normally some filtering mechanisms are employed in order to remove ``spam'' images that are not appropriate for training. In this paper, we investigate several aspects of a web-supervised system that has to be adapted to another target domain: 1.) the effect of incremental learning, 2.) the effect of spam filtering, and 3.) the behavior of particular concept classes with respect to 1.) and 2.). The experimental results provide some insights under which conditions incremental learning and spam filtering are useful.}, acmid = {2912072}, added-at = {2017-02-07T15:32:52.000+0100}, address = {New York, NY, USA}, author = {Springstein, Matthias and Ewerth, Ralph}, biburl = {https://www.bibsonomy.org/bibtex/264d129e2e3fa5a897efacb89627d095a/alexandriaproj}, booktitle = {Proceedings of the 2016 ACM on International Conference on Multimedia Retrieval}, doi = {10.1145/2911996.2912072}, interhash = {adaf35cfb79014cc03925a5561cb5a5c}, intrahash = {64d129e2e3fa5a897efacb89627d095a}, isbn = {978-1-4503-4359-6}, keywords = {alexandria}, location = {New York, New York, USA}, numpages = {4}, pages = {377--380}, publisher = {ACM}, series = {ICMR '16}, timestamp = {2017-02-07T15:32:52.000+0100}, title = {On the Effects of Spam Filtering and Incremental Learning for Web-Supervised Visual Concept Classification}, url = {http://doi.acm.org/10.1145/2911996.2912072}, year = 2016 } @inproceedings{springstein2016effects, abstract = {Deep neural networks have been successfully applied to the task of visual concept classification. However, they require a large number of training examples for learning. Although pre-trained deep neural networks are available for some domains, they usually have to be fine-tuned for an envisaged target domain. Recently, some approaches have been suggested that are aimed at incrementally (or even endlessly) learning visual concepts based on Web data. Since tags of Web images are often noisy, normally some filtering mechanisms are employed in order to remove "spam" images that are not appropriate for training. In this paper, we investigate several aspects of a web-supervised system that has to be adapted to another target domain: 1.) the effect of incremental learning, 2.) the effect of spam filtering, and 3.) the behavior of particular concept classes with respect to 1.) and 2.). The experimental results provide some insights under which conditions incremental learning and spam filtering are useful.}, added-at = {2017-01-11T21:02:01.000+0100}, author = {Springstein, Matthias and Ewerth, Ralph}, biburl = {https://www.bibsonomy.org/bibtex/2c247ea83161ed26789b996b8ac017459/ewerth}, booktitle = {ACM International Conference on Multimedia Retrieval (ICMR)}, editor = {Kender, John R. and Smith, John R. and Luo, Jiebo and Boll, Susanne and Hsu, Winston H.}, interhash = {adaf35cfb79014cc03925a5561cb5a5c}, intrahash = {c247ea83161ed26789b996b8ac017459}, keywords = {deep_learning web_supervised}, pages = {377-380}, publisher = {ACM}, timestamp = {2017-01-11T21:30:43.000+0100}, title = {On the Effects of Spam Filtering and Incremental Learning for Web-Supervised Visual Concept Classification}, url = {http://doi.acm.org/10.1145/2911996.2912072}, year = 2016 }