{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/deep-neural-networks-regularization-for","title":"Deep Neural Networks Regularization for Structured Output Prediction","arxiv_id":"1504.07550","date":"2015-04-28","proceeding":null,"authors":["Soufiane Belharbi","Romain Hérault","Clément Chatelain","Sébastien Adam"],"abstract":"A deep neural network model is a powerful framework for learning\nrepresentations. Usually, it is used to learn the relation $x \\to y$ by\nexploiting the regularities in the input $x$. In structured output prediction\nproblems, $y$ is multi-dimensional and structural relations often exist between\nthe dimensions. The motivation of this work is to learn the output dependencies\nthat may lie in the output data in order to improve the prediction accuracy.\nUnfortunately, feedforward networks are unable to exploit the relations between\nthe outputs. In order to overcome this issue, we propose in this paper a\nregularization scheme for training neural networks for these particular tasks\nusing a multi-task framework. Our scheme aims at incorporating the learning of\nthe output representation $y$ in the training process in an unsupervised\nfashion while learning the supervised mapping function $x \\to y$.\n  We evaluate our framework on a facial landmark detection problem which is a\ntypical structured output task. We show over two public challenging datasets\n(LFPW and HELEN) that our regularization scheme improves the generalization of\ndeep neural networks and accelerates their training. The use of unlabeled data\nand label-only data is also explored, showing an additional improvement of the\nresults. We provide an opensource implementation\n(https://github.com/sbelharbi/structured-output-ae) of our framework.","url_abs":"http://arxiv.org/abs/1504.07550v6","url_pdf":"http://arxiv.org/pdf/1504.07550v6.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"deep-neural-networks-regularization-for","repo_url":"https://github.com/sbelharbi/structured-output-ae","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"none","reach":null}],"tasks":[{"task_slug":"facial-landmark-detection","task_name":"Facial Landmark Detection"},{"task_slug":"prediction","task_name":"Prediction"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}