{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2025,7,30]],"date-time":"2025-07-30T15:50:04Z","timestamp":1753890604713,"version":"3.41.2"},"reference-count":27,"publisher":"Frontiers Media SA","license":[{"start":{"date-parts":[[2024,5,22]],"date-time":"2024-05-22T00:00:00Z","timestamp":1716336000000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"funder":[{"DOI":"10.13039\/100000185","name":"Defense Advanced Research Projects Agency","doi-asserted-by":"publisher","id":[{"id":"10.13039\/100000185","id-type":"DOI","asserted-by":"publisher"}]}],"content-domain":{"domain":["frontiersin.org"],"crossmark-restriction":true},"short-container-title":["Front. Comput. Sci."],"abstract":"<jats:p>Over the last decade, deep generative models have evolved to generate realistic and sharp images. The success of these models is often attributed to an extremely large number of trainable parameters and an abundance of training data, with limited or no understanding of the underlying data manifold. In this article, we explore the possibility of learning a deep generative model that is structured to better capture the underlying manifold's geometry, to effectively improve image generation while providing implicit controlled generation by design. Our approach structures the latent space into multiple disjoint representations capturing different attribute manifolds. The global representations are guided by a disentangling loss for effective attribute representation learning and a differential manifold divergence loss to learn an effective implicit generative model. Experimental results on a 3D shapes dataset demonstrate the model's ability to disentangle attributes without direct supervision and its controllable generative capabilities. These findings underscore the potential of structuring deep generative models to enhance image generation and attribute control without direct supervision with ground truth attributes signaling progress toward more sophisticated deep generative models.<\/jats:p>","DOI":"10.3389\/fcomp.2024.1274779","type":"journal-article","created":{"date-parts":[[2024,5,22]],"date-time":"2024-05-22T04:53:08Z","timestamp":1716353588000},"update-policy":"https:\/\/doi.org\/10.3389\/crossmark-policy","source":"Crossref","is-referenced-by-count":0,"title":["Orthogonality and graph divergence losses promote disentanglement in generative models"],"prefix":"10.3389","volume":"6","author":[{"given":"Ankita","family":"Shukla","sequence":"first","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Rishi","family":"Dadhich","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Rajhans","family":"Singh","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Anirudh","family":"Rayas","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Pouria","family":"Saidi","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Gautam","family":"Dasarathy","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Visar","family":"Berisha","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Pavan","family":"Turaga","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]}],"member":"1965","published-online":{"date-parts":[[2024,5,22]]},"reference":[{"key":"B1","doi-asserted-by":"publisher","first-page":"1947","DOI":"10.1109\/ITA.2018.8503149","article-title":"Emergence of invariance and disentanglement in deep representations","volume":"19","author":"Achille","year":"2018","journal-title":"J. Machine Learn. Res."},{"key":"B2","first-page":"399","article-title":"\u201cInvertible generative models for inverse problems: mitigating representation error and dataset bias,\u201d","volume-title":"International Conference on Machine Learning","author":"Asim","year":"2020"},{"key":"B3","doi-asserted-by":"publisher","first-page":"18","DOI":"10.1109\/MSP.2017.2693418","article-title":"Geometric deep learning: going beyond euclidean data","volume":"34","author":"Bronstein","year":"2017","journal-title":"IEEE Sign. Process. Mag."},{"key":"B4","unstructured":"BurgessC.\n            KimH.\n          2018"},{"key":"B5","first-page":"4229","article-title":"\u201cRecursive fr\u00e9chet mean computation on the grassmannian and its applications to computer vision,\u201d","author":"Chakraborty","year":"2015","journal-title":"IEEE International Conference on Computer Vision, ICCV 2015"},{"key":"B6","unstructured":"Role of orthogonality constraints in improving properties of deep networks for image classification\n            ChoiH.\n            SomA.\n            TuragaP. K.\n          CoRR2020"},{"key":"B7","unstructured":"\u201cLearning implicit generative models using differentiable graph tests,\u201d\n            DjolongaJ.\n            KrauseA.\n          Advances in Approximate Bayesian Inference NIPS Workshop2017"},{"key":"B8","doi-asserted-by":"publisher","first-page":"643","DOI":"10.1109\/34.927464","article-title":"From few to many: illumination cone models for face recognition under variable lighting and pose","volume":"23","author":"Georghiades","year":"2001","journal-title":"IEEE Trans. Pattern Anal. Mach. Intell."},{"key":"B9","first-page":"2672","article-title":"\u201cGenerative adversarial nets,\u201d","author":"Goodfellow","year":"2014","journal-title":"Advances in Neural Information Processing Systems 27: Annual Conference on Neural Information Processing Systems 2014, December 8-13 2014, Montreal, Quebec, Canada"},{"key":"B10","first-page":"1989","article-title":"\u201cCYCADA: cycle-consistent adversarial domain adaptation,\u201d","volume-title":"International Conference on Machine Learning","author":"Hoffman","year":"2018"},{"key":"B11","unstructured":"26492658\n            KimH.\n            MnihA.\n          2018"},{"key":"B12","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.1312.6114","article-title":"Auto-encoding variational bayes","author":"Kingma","year":"2013","journal-title":"arXiv preprint arXiv:1312.6114"},{"key":"B13","unstructured":"KlindtD. A.\n            SchottL.\n            SharmaY.\n            UstyuzhaninovI.\n            BrendelW.\n            BethgeM.\n          2021"},{"key":"B14","first-page":"1329","article-title":"\u201cLearning invariant riemannian geometric representations using deep nets,\u201d","volume-title":"IEEE International Conference on Computer Vision Workshops, ICCV Workshops 2017","author":"Lohit","year":"2017"},{"key":"B15","first-page":"4212","article-title":"\u201cDisentangled graph convolutional networks,\u201d","volume-title":"International Conference on Machine Learning","author":"Ma","year":"2019"},{"key":"B16","unstructured":"MattheyL.\n            HigginsI.\n            HassabisD.\n            LerchnerA.\n          dsprites: Disentanglement Testing Sprites Dataset2017"},{"key":"B17","first-page":"427","article-title":"\u201cDeep neural networks are easily fooled: high confidence predictions for unrecognizable images,\u201d","author":"Nguyen","year":"2015","journal-title":"IEEE Conference on Computer Vision and Pattern Recognition (CVPR)"},{"journal-title":"Greedy, Brittle, Opaque, and Shallow: The Downsides to Deep Learning","year":"2018","author":"Pontin","key":"B18"},{"key":"B19","first-page":"4288","article-title":"\u201cThe riemannian geometry of deep generative models,\u201d","volume-title":"2018 IEEE\/CVF Conference on Computer Vision and Pattern Recognition Workshops","author":"Shao","year":"2018"},{"key":"B20","unstructured":"\u201cPrOSe: product of orthogonal spheres parameterization for disentangled representation learning,\u201d88\n            ShuklaA.\n            BhagatS.\n            UppalS.\n            AnandS.\n            TuragaP. K.\n          BMVA Press30th British Machine Vision Conference 2019, BMVC 2019, Cardiff, UK, September 9-12, 20192019"},{"volume-title":"Riemannian Computing in Computer Vision, 1st Edn","year":"2015","author":"Turaga","key":"B21"},{"key":"B22","unstructured":"Visualizing data using t-SNE25792605\n            van der MaatenL.\n            HintonG.\n          J. Machine Learn. Res.92008"},{"key":"B23","first-page":"1001","article-title":"\u201cDisentangled graph collaborative filtering,\u201d","author":"Wang","year":"2020","journal-title":"Proceedings of the 43rd International ACM SIGIR Conference on Research and Development in Information Retrieval"},{"key":"B24","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.2303.07909","article-title":"Text-to-image diffusion model in generative AI: a survey","author":"Zhang","year":"2023","journal-title":"arXiv preprint arXiv:2303.07909"},{"key":"B25","first-page":"592","article-title":"\u201cIn-domain gan inversion for real image editing,\u201d","volume-title":"European Conference on Computer Vision","author":"Zhu","year":"2020"},{"key":"B26","first-page":"2223","article-title":"\u201cUnpaired image-to-image translation using cycle-consistent adversarial networks,\u201d","author":"Zhu","year":"2017","journal-title":"Proceedings of the IEEE International Conference on Computer Vision"},{"key":"B27","first-page":"12979","article-title":"\u201cContrastive learning inverts the data generating process,\u201d","author":"Zimmermann","year":"2021","journal-title":"Proceedings of the 38th International Conference on Machine Learning"}],"container-title":["Frontiers in Computer Science"],"original-title":[],"link":[{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/fcomp.2024.1274779\/full","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2024,5,22]],"date-time":"2024-05-22T04:53:19Z","timestamp":1716353599000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/fcomp.2024.1274779\/full"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2024,5,22]]},"references-count":27,"alternative-id":["10.3389\/fcomp.2024.1274779"],"URL":"https:\/\/doi.org\/10.3389\/fcomp.2024.1274779","relation":{},"ISSN":["2624-9898"],"issn-type":[{"type":"electronic","value":"2624-9898"}],"subject":[],"published":{"date-parts":[[2024,5,22]]},"article-number":"1274779"}}