Since August 2014, I am a PhD student in Prof. Manfred Claassen's Computational Biology group at ETH Zurich. I use machine learning to dissect cellular heterogeneity and study its effects in cancer and immune biology. My main focus lies on representation learning methods and their application for the analysis of high-dimensional single cell profiles, such as those obtained by mass cytometry or single cell RNA-seq. We have recently developed CellCnn, a novel method to detect phenotype-associated, possibly rare, cell subsets. CellCnn formulates this detection problem as a multiple-instance learning task and addresses it by means of a convolutional neural network.
Before, I completed a MSc in Computational Biology and Bioinformatics at ETH Zurich. I did my master thesis with Prof. Manfred Claassen, working on structure learning of intracellular signaling networks in apoptosis using undirected probabilistic graphical models. My first degree is a 5-year Diploma in Electrical and Computer Engineering from the National Technical University of Athens. For my diploma thesis, I worked under the supervision of Prof. Kostis Sagonas, contributing to the development of PropEr: a Property-based testing tool for Erlang.
Arvaniti E., & Claassen M. (2016). Sensitive detection of rare disease-associated cell subsets via representation learning. http://biorxiv.org/content/early/2016/03/31/046508
Arvaniti, E., & Claassen, M. (2014). Markov Network Structure Learning via Ensemble-of-Forests Models. Uncertainty in Artificial Intelligence (UAI).
Szklarczyk, O. M., Arvaniti, E., & van Gunsteren, W. F. (2015). Polarizable coarse‐grained models for molecular dynamics simulation of liquid cyclohexane. Journal of computational chemistry, 36(17), 1311-1321.
Current and past open-source projects
CellCnn: Representation Learning for detection of disease-associated cell subsets.
PropEr: a Property-based testing tool for the Erlang programming language.