Learning Trajectory Routing with Graph Neural Networks

Yu, Zhishui; Guo, Yuchun; Chen, Yishuai

doi:10.1145/3404687.3404701

Cited by 21 publications

(42 citation statements)

References 5 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In a recent work, DAG with NoTears [52], the authors converted the combinatorial optimization problem into a continuous one and provided an associated optimization algorithm to recover sparse DAGs. This eventually spawned research that provided improvements and also capture complex functional dependencies by introducing deep learning variants [51,50]. One such follow up work is by [53], which we found to be the close to our method in terms of function representation capacity, specifically the MLP version of the nonparametric DAG learning.…”

Section: Related Methodsmentioning

confidence: 73%

Neural Graph Revealers

Shrivastava¹,

Chajewska²

2023

Preprint

View full text Add to dashboard Cite

Sparse graph recovery methods work well where the data follows their assumptions but often they are not designed for doing downstream probabilistic queries. This limits their adoption to only identifying connections among the input variables. On the other hand, the Probabilistic Graphical Models (PGMs) assume an underlying base graph between variables and learns a distribution over them. PGM design choices are carefully made such that the inference & sampling algorithms are efficient. This brings in certain restrictions and often simplifying assumptions. In this work, we propose Neural Graph Revealers (NGRs), that are an attempt to efficiently merge the sparse graph recovery methods with PGMs into a single flow. The problem setting consists of an input data X with D features and M samples and the task is to recover a sparse graph showing connection between the features and learn a probability distribution over the D at the same time. NGRs view the neural networks as a 'glass box' or more specifically as a multitask learning framework. We introduce 'Graph-constrained path norm' that NGRs leverage to learn a graphical model that captures complex non-linear functional dependencies between the features in the form of an undirected sparse graph. Furthermore, NGRs can handle multimodal inputs like images, text, categorical data, embeddings etc. which is not straightforward to incorporate in the existing methods. We show experimental results of doing sparse graph recovery and probabilistic inference on data from Gaussian graphical models and a multimodal infant mortality dataset by Centers for Disease Control and Prevention.

show abstract

Section: Related Methodsmentioning

confidence: 73%

Neural Graph Revealers

Shrivastava¹,

Chajewska²

2023

Preprint

View full text Add to dashboard Cite

show abstract

“…This is not a realistic assumption in observational sciences. Yu et al (2019) and Xu and Xu (2021) propose methods using neural networks in the same special case where the full DAGs is identifiable, and they also consider only a fixed average graph density, and do not study the influence of sample size. Ke et al (2022) also address this special case using neural networks, and they furthermore train their model on a mixture of observational and experimental data.…”

Section: Related Workmentioning

confidence: 99%

Causal Discovery for Observational Sciences Using Supervised Machine Learning

Petersen¹,

Ramsey²,

Ekstrøm³

et al. 2023

Journal of Data Science

View full text Add to dashboard Cite

Causal inference can estimate causal effects, but unless data are collected experimentally, statistical analyses must rely on pre-specified causal models. Causal discovery algorithms are empirical methods for constructing such causal models from data. Several asymptotically correct discovery methods already exist, but they generally struggle on smaller samples. Moreover, most methods focus on very sparse causal models, which may not always be a realistic representation of real-life data generating mechanisms. Finally, while causal relationships suggested by the methods often hold true, their claims about causal non-relatedness have high error rates. This non-conservative error trade off is not ideal for observational sciences, where the resulting model is directly used to inform causal inference: A causal model with many missing causal relations entails too strong assumptions and may lead to biased effect estimates. We propose a new causal discovery method that addresses these three shortcomings: Supervised learning discovery (SLdisco). SLdisco uses supervised machine learning to obtain a mapping from observational data to equivalence classes of causal models. We evaluate SLdisco in a large simulation study based on Gaussian data and we consider several choices of model size and sample size. We find that SLdisco is more conservative, only moderately less informative and less sensitive towards sample size than existing procedures. We furthermore provide a real epidemiological data application. We use random subsampling to investigate real data performance on small samples and again find that SLdisco is less sensitive towards sample size and hence seems to better utilize the information available in small datasets.

show abstract

“…Recently, a new method called 'DAG with NOTEARS' [45] was introduced which converts the combinatorial optimization problem of DAG learning into a continuous one. This has led to the development of many follow up works, including some deep learning methods [43,44,46,24]. There are also many reviews of causal structure learning methods and structural equation models available, for instance the work in [16].…”

Section: Directed Graphsmentioning

confidence: 99%

Reconstruction of Gene Regulatory Networks using sparse graph recovery models

Shrivastava

2023

Preprint

View full text Add to dashboard Cite

There is a considerable body of work in the field of computer science on the topic of sparse graph recovery, particularly with regards to the innovative deep learning approaches that have been recently introduced. Despite this abundance of research, however, these methods are often not applied to the recovery of Gene Regulatory Networks (GRNs). This work aims to initiate this trend by highlighting the potential benefits of using these computational techniques in the recovery of GRNs from single cell RNA sequencing or bulk sequencing based gene expression data. GRNs are directed graphs that capture the direct dependence between transcription factors (TFs) and their target genes. Understanding these interactions is vital for studying the mechanisms in cell differentiation, growth and development. We categorize graph recovery methods into four main types based on the underlying formulations: Regression-based, Graphical Lasso, Markov Networks and Directed Acyclic Graphs. We selected representative methods from each category and made modifications to incorporate transcription factor information as a prior to ensure successful reconstruction of GRNs.

show abstract

Learning Trajectory Routing with Graph Neural Networks

Cited by 21 publications

References 5 publications

Neural Graph Revealers

Neural Graph Revealers

Causal Discovery for Observational Sciences Using Supervised Machine Learning

Reconstruction of Gene Regulatory Networks using sparse graph recovery models

Contact Info

Product

Resources

About