Machine Learning Researcher
Measure. Identify. Reason.
I build the foundations for machine learning to answer scientific “what if” questions: measuring the right variables, identifying stable structure, and reasoning about worlds we have never observed.
About
Final-year ELLIS PhD candidate at ISTA and MPI-IS · Google PhD Fellow · Visiting researcher at Caltech.
Science asks about worlds we never observed. What would the climate look like without human-caused warming? Answering such questions requires representations that preserve scientific variables and mechanisms that remain meaningful under intervention.
My research connects measurement validity, causal identifiability, and physical mechanisms. Together, they form a path from learning patterns in observations to reasoning about how a system would respond to change.
Research interests
Recognition
2025–2027
Machine Learning and ML Foundations
2026
Competitive Europe-wide GPU/HPC compute allocation
2024
First-author paper
2022–2023
Amazon–MPI Science Hub
Research
Three connected steps toward counterfactual reasoning for science—from trustworthy measurements to unseen worlds.
Are we measuring what matters?
A strong predictor is not automatically a valid scientific instrument. TC-Bench tests whether representations preserve physical distinctions; T-MEX examines whether learned proxies support the causal question we want to ask.
ICML 2026 · NeurIPS 2025
What structure makes recovery possible?
I study how invariance, multiple views, and dynamical structure make scientific variables identifiable. These principles connect learned representations to the underlying physical parameters and mechanisms across changing conditions.
ICLR 2025 · NeurIPS 2024
What if the world had been different?
Publications
Conference papers and research publications across causal representation learning, scientific machine learning, and probabilistic numerics.
Dingling Yao, Kahaan Gandhi, Valentin Duruisseaux, Boris Bonev, Francesco Locatello, Anima Anandkumar
Preprint · arXiv:2610.02252
Dingling Yao, Andrea Polesello, Adeel Pervez, Caroline Muller, Francesco Locatello
ICML 2026
Dingling Yao, S. Huang, R. Cadei, K. Zhang, F. Locatello
NeurIPS 2025 · Equal contribution: D. Yao and S. Huang
Dingling Yao, Dario Rancati, Riccardo Cadei, Marco Fumero, Francesco Locatello
ICLR 2025
Jiale Chen, Dingling Yao, Adeel Pervez, Dan Alistarh, Francesco Locatello
ICLR 2025
Dingling Yao, Filip Tronarp, Nathanael Bosch
ProbNum 2025 Oral
Dingling Yao, Danru Xu, Sébastien Lachapelle, Sara Magliacane, Perouz Taslakian, Georg Martius, Julius von Kügelgen, Francesco Locatello
ICLR 2024 Spotlight · Top 5%
Danru Xu, Dingling Yao, Sébastien Lachapelle, Perouz Taslakian, Julius von Kügelgen, Francesco Locatello, Sara Magliacane
ICML 2024
Talks
Selected talks and workshop presentations across causality, uncertainty, and scientific AI.
September 2026
International Institute for Applied Systems Analysis · Laxenburg, Austria
September 2026
U.S. talk series · UC San Diego, UC Santa Barbara, Broad Institute of MIT and Harvard, Columbia University, and New York University
April 2025
Institute of Computing Technology, Chinese Academy of Sciences · Remote
March 2024
Invited talk · Trieste, Italy
February 2024
Invited talk · Barbados
December 2023
Contributed talk · New Orleans, US
Teaching & mentoring
ISTA · Co-supervision
ICLR 2025
Scalable Mechanistic Neural NetworksCaltech · Mentoring
Academic Service
Workshop organizing, speaker series coordination, and reviewing across the ML research community.
2025 · Lead Organizer
Led the inaugural workshop on causality and science, with more than 300 participants.
2025–present · Organizer
Co-organizing a series on unifying representations, with more than 100 participants.
Ongoing · Reviewer
Contact
For collaborations, talks, workshop ideas, or research conversations, email is the fastest path.
