High Reward DAG

High-reward Directed Acyclic Graph (DAG) learning focuses on efficiently and accurately inferring causal relationships from data, represented as a DAG where nodes represent variables and edges represent causal influences. Current research emphasizes developing algorithms that overcome challenges like handling multiple co-existing DAGs, efficiently searching the vast space of possible DAGs (using methods like generative flow networks and momentum-based sampling), and robustly estimating parameters even with incomplete or heteroscedastic data (employing techniques such as optimal transport and concomitant linear estimation). These advancements are crucial for improving causal inference in various fields, enabling more reliable predictions and interventions in complex systems.

Papers

October 29, 2024

DAGE: DAG Query Answering via Relational Combinator with Logical Constraints
Yunjie He, Bo Xiong, Daniel Hernández, Yuqicheng Zhu, Evgeny Kharlamov, Steffen Staab
Knowledge Graph Participation Constraint Join Algorithm High Reward DAG Query Embeddings

October 18, 2024

CaTs and DAGs: Integrating Directed Acyclic Graphs with Transformers and Fully-Connected Neural Networks for Causally Constrained Predictions
Matthew J. Vowels, Mathieu Rochat, Sina Akbari
Transformer Megatron Decepticons Causal Structure Causal Direction Directed Acyclic Graph Causal Transformer Fully Connected Neural Network Fully Connected Causal Constraint High Reward DAG

June 12, 2024

Interventional Causal Discovery in a Mixture of DAGs
Burak Varıcı, Dmitriy Katz-Rogozhnikov, Dennis Wei, Prasanna Sattigeri, Ali Tajer
Causal Discovery Mixture Component Causal Graph Causal Variable Causal System High Reward DAG Causal Structure Discovery

October 9, 2023

Causal structure learning with momentum: Sampling distributions over Markov Equivalence Classes of DAGs
Moritz Schauer, Marcel Wienöbst
Causal Structure Residual Momentum Fast Sampling Markov Equivalence Class Bayesian Network Structure Reversible Markov Chain High Reward DAG

October 4, 2023

CoLiDE: Concomitant Linear DAG Estimation
Seyed Saman Saboksayr, Gonzalo Mateos, Mariano Tepper
Directed Acyclic Graph Scoring Function TaCo MatAtlas Cendol Lasso Problem High Reward DAG Non Convex Penalty

May 25, 2023

October 15, 2022

GFlowCausal: Generative Flow Networks for Causal Discovery
Wenqian Li, Yinchuan Li, Shengyu Zhu, Yunfeng Shao, Jianye Hao, Yan Pang
Active Learning Causal Discovery Causal Structure Generative Flow Network High Reward DAG

April 10, 2022

From graphs to DAGs: a low-complexity model and a scalable algorithm
Shuyu Dong, Michèle Sebag
Graph Drawing Matrix Factorization Low Rank Directed Acyclic Graph Sparse Matrix Multi Scale Graph Scalable Algorithm High Reward DAG Low Complexity Model

February 22, 2022

Causal Effect Identification in Cluster DAGs
Tara V. Anand, Adèle H. Ribeiro, Jin Tian, Elias Bareinboim
Counterfactual Inference Causal Effect Identification Causal Hierarchy High Reward DAG

December 7, 2021

FedDAG: Federated DAG Structure Learning
Erdun Gao, Junjia Chen, Li Shen, Tongliang Liu, Mingming Gong, Howard Bondell
Graph Drawing Gradient Descent Gradient Based High Reward DAG Federated Causal