- UNSW
- ...
- Our schools
- Mathematics & Statistics
- Engage with us
- Seminars
- 2023
- Contextual directed acyclic graphs
- Home
- Our school
- Study with us
- Our research
-
Student life & resources
- Undergraduate
- Honours year
- Postgraduate coursework
-
Postgraduate research
- Info for new students
- Current research students
- Postgraduate conference
- Postgraduate events
- Postgraduate student awards
- Michael Tallis PhD Research Travel Award
- Information about research theses
- Past research students
- Resources
- Entry requirements
- PhD projects
- Obtaining funding
- Application & fee information
-
Student services
- Help for postgraduate students
- Thesis guidelines
- School assessment policies
- Computing information
- Mathematics Drop-in Centre
- Consultation
- Statistics Consultation Service
- Academic advice
- Enrolment variation
- Changing tutorials
- Illness or misadventure
- Application form for existing casual tutors
- ARC grants Head of School sign off
- Computing facilities
- Choosing your major
- Student societies
- Student noticeboard
- Casual tutors
- Engage with us
- News & events
- Contact
- Home
- Our school
- Study with us
- Our research
-
Student life & resources
Postgraduate research
- Info for new students
- Current research students
- Postgraduate conference
- Postgraduate events
- Postgraduate student awards
- Michael Tallis PhD Research Travel Award
- Information about research theses
- Past research students
- Resources
- Entry requirements
- PhD projects
- Obtaining funding
- Application & fee information
Student services
- Help for postgraduate students
- Thesis guidelines
- School assessment policies
- Computing information
- Mathematics Drop-in Centre
- Consultation
- Statistics Consultation Service
- Academic advice
- Enrolment variation
- Changing tutorials
- Illness or misadventure
- Application form for existing casual tutors
- ARC grants Head of School sign off
- Computing facilities
- Choosing your major
- Engage with us
- News & events
- Contact
Abstract
Estimating the structure of directed acyclic graphs (DAGs) from observational data remains a significant challenge in machine learning. Most research in this area concentrates on learning a single DAG for the entire population. This paper considers an alternative setting where the graph structure varies across individuals based on available "contextual" features. We tackle this contextual DAG problem via a neural network that maps the contextual features to a DAG, represented as a weighted adjacency matrix. The neural network is equipped with a novel projection layer that ensures the output matrices are sparse and satisfy a recently developed characterisation of acyclicity. We devise a scalable computational framework for learning contextual DAGs and provide a convergence guarantee and an analytical gradient for backpropagating through the projection layer. Our experiments suggest that the new approach can recover the true context-specific graph where existing approaches fail.