Danru Xu

🐼

Danru Xu 徐丹茹

PhD student

About Me

Hi! I am a PhD student in the INDE lab at the University of Amsterdam (UvA) from 2023, supervised by Prof. Paul Groth and A/Prof. Sara Magliacane. My research mainly focuses on causality, causal representation learning, and multimodal data fusion.

Before starting my Ph.D., I obtained my master’s degree from the University of Melbourne with a thesis on DAG learning (supervised by A/Prof. Mingming Gong), and my bachelor’s degree from Chongqing University. I also joined an undergraduate exchange program at the National Taiwan University and a study abroad program at the University of Queensland, both are in the department of mathematics.

Interests

Causal Representation Learning

Education

MSc in Mathematics and Statistics
The University of Melbourne
BSc in Mathematics and Statistics
Chongqing University

Featured Publications

Partial Observable Causal Representation Learning

A sparsity principle for partially observable causal representation learning

Most CRL methods assume that all latent causal variables are captured in the high-dimensional observations. We instead consider a partially observed setting, in which each measurement only provides information about a subset of the underlying causal state. Here we focus on learning from unpaired observations from a dataset with an instance-dependent partial observability pattern. Our main contribution is to establish two identifiability results for this setting':' one for linear mixing functions without parametric assumptions on the underlying causal model, and one for piecewise linear mixing functions with Gaussian latent causal variables.

May 1, 2024

Partial Observable and Multi-View Causal Representation Learning

Multi-view causal representation learning with partial observability

We present a unified framework for studying the identifiability of representations learned from simultaneously observed views, such as different data modalities. We allow a partially observed setting in which each view constitutes a nonlinear mixture of a subset of underlying latent variables, which can be causally related. We prove that the information shared across all subsets of any number of views can be learned up to a smooth bijection using contrastive learning and a single encoder per view.

Jan 15, 2024

Recent Publications

Danru Xu, Dingling Yao, Sébastien Lachapelle, Perouz Taslakian, Julius Von Kügelgen, Francesco Locatello, Sara Magliacane (2024). A sparsity principle for partially observable causal representation learning. In ICML 2024.

PDF Cite Code Poster

Dingling Yao, Danru Xu, Sébastien Lachapelle, Sara Magliacane, Perouz Taslakian, Georg Martius, Julius Von Kügelgen, Francesco Locatello (2024). Multi-view causal representation learning with partial observability. In ICLR 2024 (Spotlight).

PDF Cite Code Slides Video

Danru Xu, Erdun Gao, Wei Huang, Menghan Wang, Andy Song, Mingming Gong (2022). On the Sparse DAG Structure Learning Based on Adaptive Lasso. arXiv preprint arXiv:2209.02946.

PDF Cite