{"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-autoregressive-neural-networks-for-high","title":"Deep autoregressive neural networks for high-dimensional inverse problems in groundwater contaminant source identification","arxiv_id":"1812.09444","date":"2018-12-22","proceeding":null,"authors":["Shaoxing Mo","Nicholas Zabaras","Xiaoqing Shi","Jichun Wu"],"abstract":"Identification of a groundwater contaminant source simultaneously with the\nhydraulic conductivity in highly-heterogeneous media often results in a\nhigh-dimensional inverse problem. In this study, a deep autoregressive neural\nnetwork-based surrogate method is developed for the forward model to allow us\nto solve efficiently such high-dimensional inverse problems. The surrogate is\ntrained using limited evaluations of the forward model. Since the relationship\nbetween the time-varying inputs and outputs of the forward transport model is\ncomplex, we propose an autoregressive strategy, which treats the output at the\nprevious time step as input to the network for predicting the output at the\ncurrent time step. We employ a dense convolutional encoder-decoder network\narchitecture in which the high-dimensional input and output fields of the model\nare treated as images to leverage the robust capability of convolutional\nnetworks in image-like data processing. An iterative local updating ensemble\nsmoother (ILUES) algorithm is used as the inversion framework. The proposed\nmethod is evaluated using a synthetic contaminant source identification problem\nwith 686 uncertain input parameters. Results indicate that, with relatively\nlimited training data, the deep autoregressive neural network consisting of 27\nconvolutional layers is capable of providing an accurate approximation for the\nhigh-dimensional model input-output relationship. The autoregressive strategy\nsubstantially improves the network's accuracy and computational efficiency. The\napplication of the surrogate-based ILUES in solving the inverse problem shows\nthat it can achieve accurate inversion results and predictive uncertainty\nestimates.","url_abs":"http://arxiv.org/abs/1812.09444v1","url_pdf":"http://arxiv.org/pdf/1812.09444v1.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-autoregressive-neural-networks-for-high","repo_url":"https://github.com/cics-nd/cnn-inversion","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"pytorch","reach":{"status":"ok"}}],"tasks":[{"task_slug":"computational-efficiency","task_name":"Computational Efficiency"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1812.09444","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}