{"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/linked-causal-variational-autoencoder-for","title":"Linked Causal Variational Autoencoder for Inferring Paired Spillover Effects","arxiv_id":"1808.03333","date":"2018-08-09","proceeding":null,"authors":["Vineeth Rakesh","Ruocheng Guo","Raha Moraffah","Nitin Agarwal","Huan Liu"],"abstract":"Modeling spillover effects from observational data is an important problem in\neconomics, business, and other fields of research. % It helps us infer the\ncausality between two seemingly unrelated set of events. For example, if\nconsumer spending in the United States declines, it has spillover effects on\neconomies that depend on the U.S. as their largest export market. In this\npaper, we aim to infer the causation that results in spillover effects between\npairs of entities (or units), we call this effect as \\textit{paired spillover}.\nTo achieve this, we leverage the recent developments in variational inference\nand deep learning techniques to propose a generative model called Linked Causal\nVariational Autoencoder (LCVA). Similar to variational autoencoders (VAE), LCVA\nincorporates an encoder neural network to learn the latent attributes and a\ndecoder network to reconstruct the inputs. However, unlike VAE, LCVA treats the\n\\textit{latent attributes as confounders that are assumed to affect both the\ntreatment and the outcome of units}. Specifically, given a pair of units $u$\nand $\\bar{u}$, their individual treatment and outcomes, the encoder network of\nLCVA samples the confounders by conditioning on the observed covariates of $u$,\nthe treatments of both $u$ and $\\bar{u}$ and the outcome of $u$. Once inferred,\nthe latent attributes (or confounders) of $u$ captures the spillover effect of\n$\\bar{u}$ on $u$. Using a network of users from job training dataset (LaLonde\n(1986)) and co-purchase dataset from Amazon e-commerce domain, we show that\nLCVA is significantly more robust than existing methods in capturing spillover\neffects.","url_abs":"http://arxiv.org/abs/1808.03333v4","url_pdf":"http://arxiv.org/pdf/1808.03333v4.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":"linked-causal-variational-autoencoder-for","repo_url":"https://github.com/rguo12/CIKM18-LCVA","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":{"status":"ok"}}],"tasks":[{"task_slug":"variational-inference","task_name":"Variational Inference"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1808.03333","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}