Autoencoding beyond pixels using a learned similarity metric

Autoencoding beyond pixels using a learned similarity metric. Larsen, A., B., L., Sønderby, S., K., Larochelle, H., & Winther, O. In Proceedings of The 33rd International Conference on Machine Learning, pages 1558-1566, 6, 2016. PMLR.

Paper

Autoencoding beyond pixels using a learned similarity metric [link]

Website abstract bibtex

We present an autoencoder that leverages learned representations to better measure similarities in data space. By combining a variational autoencoder (VAE) with a generative adversarial network (GAN) we can use learned feature representations in the GAN discriminator as basis for the VAE reconstruction objective. Thereby, we replace element-wise errors with feature-wise errors to better capture the data distribution while offering invariance towards e.g. translation. We apply our method to images of faces and show that it outperforms VAEs with element-wise similarity measures in terms of visual fidelity. Moreover, we show that the method learns an embedding in which high-level abstract visual features (e.g. wearing glasses) can be modified using simple arithmetic.

@inproceedings{
 title = {Autoencoding beyond pixels using a learned similarity metric},
 type = {inproceedings},
 year = {2016},
 pages = {1558-1566},
 websites = {https://proceedings.mlr.press/v48/larsen16.html},
 month = {6},
 publisher = {PMLR},
 id = {832ca8f7-22e1-3066-a37a-8dc7f7a09b99},
 created = {2022-03-28T09:45:05.728Z},
 accessed = {2022-03-26},
 file_attached = {true},
 profile_id = {235249c2-3ed4-314a-b309-b1ea0330f5d9},
 group_id = {1ff583c0-be37-34fa-9c04-73c69437d354},
 last_modified = {2022-03-30T07:22:52.165Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 citation_key = {larsenAutoencodingPixelsUsing2016},
 source_type = {inproceedings},
 notes = {ISSN: 1938-7228},
 private_publication = {false},
 abstract = {We present an autoencoder that leverages learned representations to better measure similarities in data space. By combining a variational autoencoder (VAE) with a generative adversarial network (GAN) we can use learned feature representations in the GAN discriminator as basis for the VAE reconstruction objective. Thereby, we replace element-wise errors with feature-wise errors to better capture the data distribution while offering invariance towards e.g. translation. We apply our method to images of faces and show that it outperforms VAEs with element-wise similarity measures in terms of visual fidelity. Moreover, we show that the method learns an embedding in which high-level abstract visual features (e.g. wearing glasses) can be modified using simple arithmetic.},
 bibtype = {inproceedings},
 author = {Larsen, Anders Boesen Lindbo and Sønderby, Søren Kaae and Larochelle, Hugo and Winther, Ole},
 booktitle = {Proceedings of The 33rd International Conference on Machine Learning}
}

Downloads: 0

{"_id":"XptdAzjBXPGf3p5u7","bibbaseid":"larsen-snderby-larochelle-winther-autoencodingbeyondpixelsusingalearnedsimilaritymetric-2016","author_short":["Larsen, A., B., L.","Sønderby, S., K.","Larochelle, H.","Winther, O."],"bibdata":{"title":"Autoencoding beyond pixels using a learned similarity metric","type":"inproceedings","year":"2016","pages":"1558-1566","websites":"https://proceedings.mlr.press/v48/larsen16.html","month":"6","publisher":"PMLR","id":"832ca8f7-22e1-3066-a37a-8dc7f7a09b99","created":"2022-03-28T09:45:05.728Z","accessed":"2022-03-26","file_attached":"true","profile_id":"235249c2-3ed4-314a-b309-b1ea0330f5d9","group_id":"1ff583c0-be37-34fa-9c04-73c69437d354","last_modified":"2022-03-30T07:22:52.165Z","read":false,"starred":false,"authored":false,"confirmed":"true","hidden":false,"citation_key":"larsenAutoencodingPixelsUsing2016","source_type":"inproceedings","notes":"ISSN: 1938-7228","private_publication":false,"abstract":"We present an autoencoder that leverages learned representations to better measure similarities in data space. By combining a variational autoencoder (VAE) with a generative adversarial network (GAN) we can use learned feature representations in the GAN discriminator as basis for the VAE reconstruction objective. Thereby, we replace element-wise errors with feature-wise errors to better capture the data distribution while offering invariance towards e.g. translation. We apply our method to images of faces and show that it outperforms VAEs with element-wise similarity measures in terms of visual fidelity. Moreover, we show that the method learns an embedding in which high-level abstract visual features (e.g. wearing glasses) can be modified using simple arithmetic.","bibtype":"inproceedings","author":"Larsen, Anders Boesen Lindbo and Sønderby, Søren Kaae and Larochelle, Hugo and Winther, Ole","booktitle":"Proceedings of The 33rd International Conference on Machine Learning","bibtex":"@inproceedings{\n title = {Autoencoding beyond pixels using a learned similarity metric},\n type = {inproceedings},\n year = {2016},\n pages = {1558-1566},\n websites = {https://proceedings.mlr.press/v48/larsen16.html},\n month = {6},\n publisher = {PMLR},\n id = {832ca8f7-22e1-3066-a37a-8dc7f7a09b99},\n created = {2022-03-28T09:45:05.728Z},\n accessed = {2022-03-26},\n file_attached = {true},\n profile_id = {235249c2-3ed4-314a-b309-b1ea0330f5d9},\n group_id = {1ff583c0-be37-34fa-9c04-73c69437d354},\n last_modified = {2022-03-30T07:22:52.165Z},\n read = {false},\n starred = {false},\n authored = {false},\n confirmed = {true},\n hidden = {false},\n citation_key = {larsenAutoencodingPixelsUsing2016},\n source_type = {inproceedings},\n notes = {ISSN: 1938-7228},\n private_publication = {false},\n abstract = {We present an autoencoder that leverages learned representations to better measure similarities in data space. By combining a variational autoencoder (VAE) with a generative adversarial network (GAN) we can use learned feature representations in the GAN discriminator as basis for the VAE reconstruction objective. Thereby, we replace element-wise errors with feature-wise errors to better capture the data distribution while offering invariance towards e.g. translation. We apply our method to images of faces and show that it outperforms VAEs with element-wise similarity measures in terms of visual fidelity. Moreover, we show that the method learns an embedding in which high-level abstract visual features (e.g. wearing glasses) can be modified using simple arithmetic.},\n bibtype = {inproceedings},\n author = {Larsen, Anders Boesen Lindbo and Sønderby, Søren Kaae and Larochelle, Hugo and Winther, Ole},\n booktitle = {Proceedings of The 33rd International Conference on Machine Learning}\n}","author_short":["Larsen, A., B., L.","Sønderby, S., K.","Larochelle, H.","Winther, O."],"urls":{"Paper":"https://bibbase.org/service/mendeley/bfbbf840-4c42-3914-a463-19024f50b30c/file/2bccacf0-eb26-11c7-4bfc-8365ec6a6822/Larsen_et_al___2016___Autoencoding_beyond_pixels_using_a_learned_similar.pdf.pdf","Website":"https://proceedings.mlr.press/v48/larsen16.html"},"biburl":"https://bibbase.org/service/mendeley/bfbbf840-4c42-3914-a463-19024f50b30c","bibbaseid":"larsen-snderby-larochelle-winther-autoencodingbeyondpixelsusingalearnedsimilaritymetric-2016","role":"author","metadata":{"authorlinks":{}}},"bibtype":"inproceedings","biburl":"https://bibbase.org/service/mendeley/bfbbf840-4c42-3914-a463-19024f50b30c","dataSources":["nZHrFJKyxKKDaWYM8","2252seNhipfTmjEBQ"],"keywords":[],"search_terms":["autoencoding","beyond","pixels","using","learned","similarity","metric","larsen","sønderby","larochelle","winther"],"title":"Autoencoding beyond pixels using a learned similarity metric","year":2016}