Michael Wieck-Sosa

I am very lucky to be advised by Aaditya Ramdas and Cosma Shalizi.

Here is my resume. You can reach me at mwiecksosa AT cmu DOT edu.

My research interests are at the intersection of:

Dependent data (high-dimensional time series, spatiotemporal data, time-varying networks, nonstationary, nonlinear)
Causality (causal discovery, invariant causal prediction, causal representation learning, conditional independence testing)
Machine learning (neural networks, transformers, foundation models, large language models)
Likelihood-free methods (estimation, confidence sets, goodness-of-fit testing, random feature methods, change-point detection)

Research

Papers

Estimating dynamic models by matching random features (with Cosma Shalizi). Preprint. Slides. Code.
Dynamic models with p parameters are identified by 2p+1 random features (with Cosma Shalizi). Preprint.
Conditional independence testing with a single realization of a multivariate nonstationary nonlinear time series (with Michel F. C. Haddad and Aaditya Ramdas). Preprint. Slides. Code.

Current projects

Invariant causal prediction for multimodal nonstationary time series with transformers: Applications to text sequences and LLMs (with Aaditya Ramdas).
Test-time adaptation for deep generative dynamic models via random embeddings under possible temporal distribution shifts (with Cosma Shalizi).
Simulation-based inference through random features (with Cosma Shalizi).
How foundation models learn temporal dependencies in practice: A study of Chronos-2 and TabPFN-3 (with Tom Zhang* and Chad Schafer).

* Denotes students advised.

News

December 2026: I’ll be giving an invited talk at CFE-CMStatistics 2026.
June 2026: I’ll be giving an invited talk at IWSM 2026.
April 2026: I was awarded the DeGroot-Goel Fellowship for 2026 by the CMU Statistics & Data Science Department faculty.
April 2026: I successfully proposed my thesis.