iiinova — where your digital self keeps evolving

Danru Xu

@danruxu.bsky.social

PhD student in the INtelligent Data Engineering lab at the University of Amsterdam, working in multimodal #causal #representation learning. danrux.github.io

Posts Replies Media

Reposted by Danru Xu

AMLab @amlab.bsky.social · 13/12/2024

Professor Imbens also had a mentoring session with our PhD students actively working on causality, discussing their ideas and the potential impact of their applications! 👨‍🔬👩‍🔬 @matyasch.bsky.social @roelhulsman.bsky.social @rmassidda.it @danruxu.bsky.social 🔥

1134

Danru Xu @danruxu.bsky.social · 22/07/2024

📝👥 Joint work with a group of great researchers @Dingling_Yao @seblachap @PerouzT @JKugelgen @FrancescoLocat8 @saramagliacane

000

Danru Xu @danruxu.bsky.social · 22/07/2024

3️⃣Finally, we propose two methods that implement these theoretical results and validate their effectiveness with experiments on simulated data and image benchmarks.

000

Danru Xu @danruxu.bsky.social · 22/07/2024

2️⃣We introduce two theoretical results for identifying causal variables up to permutation and element-wise transformation under partial observability. Both results leverage a sparsity constraint.

000

Danru Xu @danruxu.bsky.social · 22/07/2024

1️⃣ We formalize the Unpaired Partial Observations setting for causal representation learning, where each partial observation captures only a subset of causal variables.

000

Danru Xu @danruxu.bsky.social · 22/07/2024

☀️ ICML2024 ☀️ Tuesday 11:30-13:00 #2201 ❓Do you want to identify high-level causal variables from perceptual data under partial observable setup? 💡Here we focus on learning from unpaired observations from a dataset with an instance-dependent partial observability pattern.

110