analiza aiqbench
This paper introduces Neuro-Causal Factor Analysis (NCFA), a nonparametric framework that integrates causal structure learning with deep generative models to improve latent variable recovery and model sparsity. By enforcing Markov factorization constraints on learned directed graphs, the method offers enhanced interpretability and lower complexity compared to standard factor analysis or variational autoencoders. The work represents a significant step forward in making unsupervised representation learning more robust and causally grounded for scientific applications.
Neuro-Causal Factor Analysis
arXiv:2305.19802v2 Announce Type: replace-cross Abstract: Factor analysis (FA) is a statistical method for explaining how mutually dependent observed variables can be represented in terms of mutually independent latent factors, and it is widely used in the psychological, biological, and physical sciences. We revisit this classic method from the perspective of recent advances in causal structure learning and deep generative models, introducing a framework for Neuro-Causal Factor Analysis (NCFA). Our approach is fully nonparametric: it learns a directed graph between latent and observed variables, and then fits a deep generative model constrained to respect the Markov factorization of the graph. Empirically, on synthetic and real data, NCFA attains better reconstruction error compared to standard FA and better latent distribution recovery compared to a standard variational autoencoder, all with the advantages of sparser architecture, lower model complexity, and causal interpretability.
Abstract:Factor analysis (FA) is a statistical method for explaining how mutually dependent observed variables can be represented in terms of mutually independent latent factors, and it is widely used in the psychological, biological, and physical sciences. We revisit this classic method from the perspective of recent advances in causal structure learning and deep generative models, introducing a framework for Neuro-Causal Factor Analysis (NCFA). Our approach is fully nonparametric: it learns a directed graph between latent and observed variables, and then fits a deep generative model constrained to respect the Markov factorization of the graph. Empirically, on synthetic and real data, NCFA attains better reconstruction error compared to standard FA and better latent distribution recovery compared to a standard variational autoencoder, all with the advantages of sparser architecture, lower model complexity, and causal interpretability.
Machine Learning (stat.ML); Machine Learning (cs.LG)
arXiv:2305.19802 [stat.ML]
(or arXiv:2305.19802v2 [stat.ML] for this version)
https://doi.org/10.48550/arXiv.2305.19802 Focus to learn more arXiv-issued DOI via DataCite
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 228 cuvinte
Subiecte conexe
Autonomous mechanistic discovery of colorectal cancer vulnerabilities via multi-scale AI swarms
arXiv · 21 iul. 2026
Improving CoT Monitorability of Evaluation Awareness via Verbalization Training
LessWrong · 2 oct. 2026
From Correlation to Causation in Lane Change Prediction for Automated Driving: A Causal Explanation Framework
arXiv · 16 iun. 2026
Uncovering Insights of Compound Flooding with Data-Driven AI
arXiv · 5 iun. 2026