myrlm (ατ) isi (δοτ) edu   (pardon the Greek characters!)

Myrl Marmarelis

I try to make sense of causal models in machine learning. Currently, I am spearheading a cross-disciplinary collaboration on single-cell transcriptomics with clinical relevance. We are tackling the high dimensionality of this increasingly popular sensing modality. In another multi-university initiative, I am in charge of forecasting the outcomes of complex scenarios in environments with extreme uncertainty.

I am pursuing my doctorate under Professors Greg Ver Steeg and Aram Galstyan at the USC Information Sciences Institute. My contributions focus on the unwieldy contexts for statistical inference that demand efficient, robust, and thoughtful methods. In particular, I care about scenarios where our observations are not enough to tease out the intricacies of complicated interactions. I tailor my data-driven approaches to social, physiological, and economic systems.


M. G. Marmarelis, G. Ver Steeg, N. Jahanshad, and A. Galstyan, Bounding the Effects of Continuous Treatments for Hidden Confounders, NeurIPS 2022 CML4Impact Workshop.

M. G. Marmarelis, G. Ver Steeg, and A. Galstyan, A Metric Space for Point Process Excitations, Journal of Artificial Intelligence Research 73 (2022) 1323–1353.

M. G. Marmarelis and R. G. Ghanem, Data-driven Stochastic Optimization on Manifolds for Additive Manufacturing, Computational Materials Science 181 (2020) 109750.


I am actively seeking avenues for making a mark in our battle against climate change, either through a sociopolitical or an economic lens. Feel free to reach out with your wacky ponderations.

My face.

My entrepreneurial aspirations lie in the field of promoting long-term wellness and health through novel uses of comfortable noninvasive instruments.

Past Experience

During my undergraduate studies, I developed a technique for data-driven stochastic optimization in the context of additive manufacturing with the supervision of Professor Roger Ghanem. I also spent two summers at Bloomberg LP as a machine learning and software engineering intern, and worked part-time for a semester at Research Affiliates, LLC alongside the head of the Investment Systems team.


   09/12/2021   Toning down polarization in elections.

   10/25/2020   Exponential smoothing, coupled with a primer on Bayesian inference.

Open Source

RankedChoices.jl --- A Julia package to facilitate analysis of ranked preferences.

rolling-quantiles --- A Python package to quickly stream rolling quantiles via a backend written in C.

A Logistic Point Process with Modulated Excitations --- Associated preprint coming soon!


Check out my old website, a remnant of my aspirations for quantitative freelancing.

My undergraduate endeavors were marked by oft-unpublishable ambition.

I also co-hosted USC's first data-science hackathon for undergraduates in 2019.