Learning visual biases from human imagination
Although the human visual system can recognize many concepts under challengingconditions, it still has some biases. In this paper, we investigate whether wecan extract these biases and transfer them into a machine recognition system.We introduce a novel method that, inspired by well-known tools in h...
| Main Authors: | , , , |
|---|---|
| Other Authors: | |
| Format: | Article |
| Language: | en_US |
| Published: |
Neural Information Processing Systems Foundation
2018
|
| Online Access: | http://hdl.handle.net/1721.1/113408 https://orcid.org/0000-0001-5676-2387 https://orcid.org/0000-0003-4915-0256 |
| _version_ | 1826194875993817088 |
|---|---|
| author | Vondrick, Carl Martin Pirsiavash, Hamed Oliva, Aude Torralba, Antonio |
| author2 | Massachusetts Institute of Technology. Computer Science and Artificial Intelligence Laboratory |
| author_facet | Massachusetts Institute of Technology. Computer Science and Artificial Intelligence Laboratory Vondrick, Carl Martin Pirsiavash, Hamed Oliva, Aude Torralba, Antonio |
| author_sort | Vondrick, Carl Martin |
| collection | MIT |
| description | Although the human visual system can recognize many concepts under challengingconditions, it still has some biases. In this paper, we investigate whether wecan extract these biases and transfer them into a machine recognition system.We introduce a novel method that, inspired by well-known tools in humanpsychophysics, estimates the biases that the human visual system might use forrecognition, but in computer vision feature spaces. Our experiments aresurprising, and suggest that classifiers from the human visual system can betransferred into a machine with some success. Since these classifiers seem tocapture favorable biases in the human visual system, we further present an SVMformulation that constrains the orientation of the SVM hyperplane to agree withthe bias from human visual system. Our results suggest that transferring thishuman bias into machines may help object recognition systems generalize acrossdatasets and perform better when very little training data is available. |
| first_indexed | 2024-09-23T10:03:34Z |
| format | Article |
| id | mit-1721.1/113408 |
| institution | Massachusetts Institute of Technology |
| language | en_US |
| last_indexed | 2024-09-23T10:03:34Z |
| publishDate | 2018 |
| publisher | Neural Information Processing Systems Foundation |
| record_format | dspace |
| spelling | mit-1721.1/1134082022-09-26T15:27:26Z Learning visual biases from human imagination Vondrick, Carl Martin Pirsiavash, Hamed Oliva, Aude Torralba, Antonio Massachusetts Institute of Technology. Computer Science and Artificial Intelligence Laboratory Massachusetts Institute of Technology. Department of Electrical Engineering and Computer Science Vondrick, Carl Martin Pirsiavash, Hamed Oliva, Aude Torralba, Antonio Although the human visual system can recognize many concepts under challengingconditions, it still has some biases. In this paper, we investigate whether wecan extract these biases and transfer them into a machine recognition system.We introduce a novel method that, inspired by well-known tools in humanpsychophysics, estimates the biases that the human visual system might use forrecognition, but in computer vision feature spaces. Our experiments aresurprising, and suggest that classifiers from the human visual system can betransferred into a machine with some success. Since these classifiers seem tocapture favorable biases in the human visual system, we further present an SVMformulation that constrains the orientation of the SVM hyperplane to agree withthe bias from human visual system. Our results suggest that transferring thishuman bias into machines may help object recognition systems generalize acrossdatasets and perform better when very little training data is available. United States. Office of Naval Research. Multidisciplinary University Research Initiative (N000141010933) Google (Firm) (Research Award) Google (Firm) (Ph.D. Fellowship) 2018-02-05T15:01:33Z 2018-02-05T15:01:33Z 2015-12 Article http://purl.org/eprint/type/ConferencePaper http://hdl.handle.net/1721.1/113408 Vondrick, Carl et al. "Learning visual biases from human imagination." Advances in Neural Information Processing Systems 28 (NIPS 2015), 7-12 December, 2015, Montreal, Canada, Neural Information Processing Systems Foundation, 2015. https://orcid.org/0000-0001-5676-2387 https://orcid.org/0000-0003-4915-0256 en_US https://papers.nips.cc/paper/5781-learning-visual-biases-from-human-imagination Advances in Neural Information Processing Systems 28 (NIPS 2015) Article is made available in accordance with the publisher's policy and may be subject to US copyright law. Please refer to the publisher's site for terms of use. application/pdf Neural Information Processing Systems Foundation Neural Information Processing Systems (NIPS) |
| spellingShingle | Vondrick, Carl Martin Pirsiavash, Hamed Oliva, Aude Torralba, Antonio Learning visual biases from human imagination |
| title | Learning visual biases from human imagination |
| title_full | Learning visual biases from human imagination |
| title_fullStr | Learning visual biases from human imagination |
| title_full_unstemmed | Learning visual biases from human imagination |
| title_short | Learning visual biases from human imagination |
| title_sort | learning visual biases from human imagination |
| url | http://hdl.handle.net/1721.1/113408 https://orcid.org/0000-0001-5676-2387 https://orcid.org/0000-0003-4915-0256 |
| work_keys_str_mv | AT vondrickcarlmartin learningvisualbiasesfromhumanimagination AT pirsiavashhamed learningvisualbiasesfromhumanimagination AT olivaaude learningvisualbiasesfromhumanimagination AT torralbaantonio learningvisualbiasesfromhumanimagination |