Causal Representation Meets Stochastic Modeling under Generic Geometry

Jiaxu Ren, Yixin Wang, and Biwei Huang · 2026

Read Causal Representation Meets Stochastic Modeling under Generic Geometry by Jiaxu Ren, Yixin Wang, and Biwei Huang (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2602-05033.txt.

Description

Causal representation learning tries to recover the hidden variables behind raw observations, along with the arrows between them. Jiaxu Ren and 2 co-authors extend that program to continuous time in this 2026 machine learning paper. Many real systems, from firing neurons to accumulating mutations, are point processes that evolve continuously, and earlier identifiable methods handled only snapshots or discrete steps.

The authors study when the hidden process can be identified at all, by studying the shape of the parameter space. They then build MUTATE, a variational autoencoder with a time-adaptive transition module that infers the hidden stochastic dynamics. In simulated and empirical studies, the system answered concrete scientific questions, including how mutations accumulate in genomics and what triggers neuron spikes under changing conditions.

Catalog record

CREATORJiaxu Ren, Yixin Wang, and Biwei Huang
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE145 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2602-05033

Browse the full library · About the Tower of Babel