PhD thesis at Institut Curie
dimension reduction of high-dimensional data using unsupervised machine learning
30 January 2016
Skills
- Machine learning
- data visualization
- R
- public speaking
- scientific writing
- healthcare
Summary of the PhD thesis achievements
- title: Unsupervised deconvolution of cell and environment specific signals and thier interractions in biological samples
- pushed the borders of knowledge in statistical modelling of multidimensional genetical data
- used blind source separation methods, unsupervised machine learning
- analysed big volume of data (>100 datasets, (~30 000 samples by ~20 000 genes))
- applied data clustering and dimension reduction
- worked with and publish open source code on git platforms and docker
- developed an R package for transcriptomic data analysis
- contributed > 3 to scientific publications as data analyst and 2 as a first author (publication list: https://orcid.org/0000-0002-5244-0708)
- wrote scientific publications with LaTeX/.Rmd
- worked in interdisciplinary team of 15 people
- interacted with experimentalists (bench scientists, medical doctors) and theoreticians (mathematicians, IT, physicists)
- won 2nd place of professional pitch competition, ABG
- won 3rd place in scientific pitch competition
- arrived 8th/208 in Capsim Simulation business competition as a part of a student team
- delivered scientific presentations in front of >100 scientific experts
- delivered pitches and presentations to general public (>1000 people)
- created graphical supports with vector graphics
- taught in Statistics, Informatics & Mathematics division at Pharma faculty
- completed courses on: Big data, Machine Learning, Analytics, Communication, Certificate in Business & Administration: Business & Management & Strategy & Finance , Technologies & Public Policy