Romain Lopez
 |
Ph.D. student, Electrical Engineering & Computer Sciences, UC Berkeley Affiliations: Berkeley Artificial Intelligence Research & Center for Computational Biology |
Research Interests
My interests include mathematical statistics, applied statistics and applications to machine learning in biology. More recently, I am focusing on probabilistic modeling of single-cell RNA sequencing data. Its ecosystem includes Bayesian modeling, variational inference, and more precisely approximate posterior inference with neural networks. I am the lead contributor to single-cell Variational Inference (scVI).
We offer opportunities for MSc students anywhere in the world to apply for a funded visiting research position at UC Berkeley to work on projects related to scVI
Example of an assignment I proposed in the past to select the students.
Publications
Preprints
Cost-Effective Incentive Allocation via Structured Counterfactual Inference
Romain Lopez, Chenchen Li, Xiang Yan, Junwu Xiong, Michael I. Jordan, Yuan Qi and Le Song
Submitted, 2019 [preprint]
Harmonization and Annotation of Single-cell Transcriptomics data with Deep Generative Models
Chenling Xu*, Romain Lopez*, Edouard Mehlman*, Jeffrey Regier, Michael I. Jordan and Nir Yosef
In review, 2019 [preprint]
Published papers
Deep Generative Modeling for Single-cell Transcriptomics
Romain Lopez, Jeffrey Regier, Michael B. Cole, Michael I. Jordan, and Nir Yosef
Nature Methods, 2018 [journal, preprint, News & Views, BAIR Blog, NIPS MLCB, BayLearn Oral]
Information Constraints on Auto-Encoding Variational Bayes
Romain Lopez, Jeffrey Regier, Michael I. Jordan and Nir Yosef
Advances in Neural Information Processing Systems, 2018 [proc., preprint]
Scrublet: computational identification of cell doublets in single-cell transcriptomic data
Samuel L. Wolock, Romain Lopez and Allon M. Klein
Cell Systems (to appear, 2018 [preprint])
* These authors contributed equally to this work.
About me
I grew up in Bedarieux, a small town in southern France, where I spent my time riding motorcycles, learning and teaching Judo as well as thinking about quantitative models for evaluating the temperature of the different planets of the solar system. I then followed the traditional French system of classes preparatoires in Toulouse at lycée Pierre-de-Fermat. After that, I worked with Laurent Massoulie (Optimal stochastic control on random graphs) at Ecole polytechnique (France) where I received the Diplome d'Ingenieur (2016). After several real-world experiences at Cardiologs (Time-series classification) and in Systems Biology, HMS with Allon Klein (Sparse Matrix Factorization), I decided to maintain this versatile orientation between theory and applications. Recently, I also worked with Le Song at Ant Financial in Hangzhou on Counterfactual Inference and Offline policy optimization.