Filters








7,545 Hits in 3.0 sec

Causal Structure Learning: a Combinatorial Perspective [article]

Chandler Squires, Caroline Uhler
2022 arXiv   pre-print
We devote special attention to two fundamental combinatorial aspects of causal structure learning. First, we discuss the structure of the search space over causal graphs.  ...  In this review, we discuss approaches for learning causal structure from data, also called causal discovery.  ...  Center at the Broad Institute, and a Simons Investigator Award.  ... 
arXiv:2206.01152v1 fatcat:k7bha6htu5cwlmhh2haquuja7m

A Combinatorial Perspective on Transfer Learning [article]

Jianan Wang, Eren Sezener, David Budden, Marcus Hutter, Joel Veness
2020 arXiv   pre-print
to contemporary deep learning techniques use a modular and local learning mechanism.  ...  We demonstrate that this system exhibits a number of desirable continual learning properties: robustness to catastrophic forgetting, no negative transfer and increasing levels of positive transfer as more  ...  A Combinatorial Perspective on Transfer Learning Jianan Wang Eren Sezener David Budden Marcus  ... 
arXiv:2010.12268v1 fatcat:krjfaqo4wfgbnozehd2fwwni5u

Multiscale Causal Structure Learning [article]

Gabriele D'Acunto, Paolo Di Lorenzo, Sergio Barbarossa
2022 arXiv   pre-print
This paper exposes a novel method, named Multiscale-Causal Structure Learning (MS-CASTLE), to estimate the structure of linear causal relationships occurring at different time scales.  ...  The inference of causal structures from observed data plays a key role in unveiling the underlying dynamics of the system.  ...  The causal structure learning problem can be formalized as follows.  ... 
arXiv:2207.07908v1 fatcat:md5m2pzt4rdo5pmsxdf5xauljy

Structural Agnostic Modeling: Adversarial Learning of Causal Graphs [article]

Diviyan Kalainathan, Olivier Goudet, Isabelle Guyon, David Lopez-Paz, Michèle Sebag
2022 arXiv   pre-print
A new causal discovery method, Structural Agnostic Modeling (SAM), is presented in this paper.  ...  A learning criterion combining distribution estimation, sparsity and acyclicity constraints is used to enforce the optimization of the graph structure and parameters through stochastic gradient descent  ...  The critical combinatorial optimization problem at the core of (causal) graph learning thus is tackled through a single continuous optimization problem.  ... 
arXiv:1803.04929v5 fatcat:ynkjkpqj6vdvplnkzfbjgntt24

Amortized Inference for Causal Structure Learning [article]

Lars Lorch, Scott Sussex, Jonas Rothfuss, Andreas Krause, Bernhard Schölkopf
2022 arXiv   pre-print
Learning causal structure poses a combinatorial search problem that typically involves evaluating structures using a score or independence test.  ...  In this work, we propose to amortize the process of causal structure learning.  ...  From a generative perspective, the observations D = {x 1 , . . . , x n } ∼ p(D) are produced by sampling from a distribution over causal structures p(G) and then obtaining realizations of a data-generating  ... 
arXiv:2205.12934v1 fatcat:o7jwjhf4vrhehkchihc4yclala

Low Rank Directed Acyclic Graphs and Causal Structure Learning [article]

Zhuangyan Fang, Shengyu Zhu, Jiji Zhang, Yue Liu, Zhitang Chen, Yangbo He
2020 arXiv   pre-print
In particular, the recent formulation of structure learning as a continuous optimization problem proved to have considerable advantages over the traditional combinatorial formulation, but the performance  ...  Despite several important advances in recent years, learning causal structures represented by directed acyclic graphs (DAGs) remains a challenging task in high dimensional settings when the graphs to be  ...  Consider the DAG shown in Application to Causal Structure Learning We now apply the low rank assumption to causal structure learning.  ... 
arXiv:2006.05691v1 fatcat:b5cwrypihfhmpoqbjv4jywfvfq

Invariant Structure Learning for Better Generalization and Causal Explainability [article]

Yunhao Ge, Sercan Ö. Arik, Jinsung Yoon, Ao Xu, Laurent Itti, Tomas Pfister
2022 arXiv   pre-print
We propose a novel framework, Invariant Structure Learning (ISL), that is designed to improve causal structure discovery by utilizing generalization as an indication.  ...  Furthermore, we extend ISL to a self-supervised learning setting where accurate causal structure discovery does not rely on any labels.  ...  To learn the representation of P a(Y ), g() should follow the causal structure of Y .  ... 
arXiv:2206.06469v1 fatcat:c5rtvroxxfgrrp5vgmdtbxmym4

Optimal experimental design via Bayesian optimization: active causal structure learning for Gaussian process networks [article]

Julius von Kügelgen, Paul K Rubenstein, Bernhard Schölkopf, Adrian Weller
2019 arXiv   pre-print
about the underlying causal structure.  ...  Starting from few observational measurements, we follow a Bayesian active learning approach to perform those experiments which, in expectation with respect to the current model, are maximally informative  ...  Active Bayesian causal discovery As opposed to causal discovery methods from observational data [5, 6] , our setting differs in that we aim to actively learn causal structure and functional relationships  ... 
arXiv:1910.03962v1 fatcat:qbmzxibekngp5iq5xktgvuie7m

Structural Causal Bandits: Where to Intervene?

Sanghack Lee, Elias Bareinboim
2018 Neural Information Processing Systems  
We leverage this characterization to build a new algorithm that takes as input a causal structure and finds a minimal, sound, and complete set of qualified arms that an agent should play to maximize its  ...  For example, a linear (or combinatorial) bandit imposes that an action x t 2 R d (or {0, 1} d ) at a time step t incurs a cost '> t x t , where 't is a loss vector chosen by, e.g., an adversary.  ...  (We note that the causal structure can easily be learned in a typical MAB setting since the agent always has interventional capabilities.)  ... 
dblp:conf/nips/LeeB18 fatcat:ziihmwac75cexjzvo3mnomddsy

Can Linear Programs Have Adversarial Examples? A Causal Perspective [article]

Matej Zečević and Devendra Singh Dhami and Kristian Kersting
2022 arXiv   pre-print
Characteristically, we show the direct influence of the Structural Causal Model (SCM) onto the subsequent LP optimization, which ultimately exposes a notion of confounding in LPs (inherited by said SCM  ...  Pearlian notion of Causality.  ...  Here causality and its Structural Causal Model come into play.  ... 
arXiv:2105.12697v5 fatcat:hpsz4fqtvnagpgmxtt3e57ixdy

Structure induction in diagnostic causal reasoning

Björn Meder, Ralf Mayrhofer, Michael R. Waldmann
2014 Psychological review  
Our structure induction model of diagnostic reasoning takes into account the uncertainty regarding the underlying causal structure.  ...  We argue against this assumption, as it neglects alternative causal structures that may have generated the sample data.  ...  This account is based on the idea that causal learning and inference are guided by general systematic assumptions about the structure of the (causal) environment, which entails a preference for fewer (  ... 
doi:10.1037/a0035944 pmid:25090421 fatcat:tgeurmarkzbc5dnt26fqecnmpq

Quantifying intrinsic causal contributions via structure preserving interventions [article]

Dominik Janzing, Patrick Blöbaum, Lenon Minorics, Philipp Faller, Atalanti Mastakouri
2021 arXiv   pre-print
To interpret the intrinsic information as a causal contribution, we consider 'structure-preserving interventions' that randomize each node in a way that mimics the usual dependence on the parents and do  ...  We propose a new notion of causal contribution which describes the 'intrinsic' part of the contribution of a node on a target node in a DAG.  ...  How To Learn The Structural Equations Since the FCM does not uniquely follow from the observed causal conditional P Xj |P Aj , it is natural to ask where it comes from.  ... 
arXiv:2007.00714v3 fatcat:jzippbc7pneubd4l2u743rncce

Cognitive novelties, informational form, and structural-causal explanations

Andrew Buskell
2020 Synthese  
structural-causal explanations.  ...  Drawing on recent work from Cecilia Heyes, and developing a case study around a novel mathematical capacity, I demonstrate how structural-causal explanations can contribute to the niche construction approach  ...  as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made.  ... 
doi:10.1007/s11229-020-02585-4 pmid:34759435 pmcid:PMC8570306 fatcat:dypiduwijbfftp2hc76li6wnm4

ABCD-Strategy: Budgeted Experimental Design for Targeted Causal Structure Discovery [article]

Raj Agrawal, Chandler Squires, Karren Yang, Karthik Shanmugam, Caroline Uhler
2019 arXiv   pre-print
Determining the causal structure of a set of variables is critical for both scientific inquiry and decision-making. However, this is often challenging in practice due to limited interventional data.  ...  That is, we assume the experimenter is interested in learning some function of the unknown graph (e.g., all descendants of a target node) subject to design constraints such as limits on the number of samples  ...  Uhler was partially supported by NSF (DMS-1651995), ONR (N00014-17-1-2147 and N00014-18-1-2765), IBM, and a Sloan Fellowship.  ... 
arXiv:1902.10347v1 fatcat:jymndt3qlfamdfdxavciiigh5q

Minimal I-MAP MCMC for Scalable Structure Discovery in Causal DAG Models [article]

Raj Agrawal and Tamara Broderick and Caroline Uhler
2018 arXiv   pre-print
Learning a Bayesian network (BN) from data can be useful for decision-making or discovering causal relationships.  ...  The resulting uncertainty about the underlying network as well as the desire to incorporate prior information recommend a Bayesian approach to learning the BN, but the highly combinatorial structure of  ...  Caroline Uhler was supported in part by NSF (DMS-1651995), ONR (N00014-17-1-2147), and a Sloan Fellowship.  ... 
arXiv:1803.05554v3 fatcat:chf4qviitzenfpm6x66jav2mrq
« Previous Showing results 1 — 15 out of 7,545 results