I work on various topics including statistics, machine learning and bio-informatics.
Multivariate analysis in high dimensional setting: dimension reduction methods including compression and variable selection. Matrix factorization: generalization of Principal Component Analysis (PCA) to non-Gaussian data, especially count data (with over-dispersion and drop-outs/zero-inflation). Supervised approaches for regression and classification using sparse Partial Least Square (sparse PLS).
Application to genomics: analyses of gene expression profiles from high-throughput sequecing data and single-cell data.
Computer vision: Convolutional Kernel Network (CKN) for image embedding.
Development and maintenance of various toolboxes for bio-statistics, machine learning:
plsgenomics (contribution and maintenance): supervised methods for dimension reduction in classification and regression framework (in particular PLS-based routines for genomic data analyses).
pCMF (full development): probabilistic count matrix factorization for single cell transcriptomic data analyses (dimension reduction, visualization).
SPAMS (maintenance): optimization toolbox for sparse estimation, implementing algorithms that solve machine learning and signal processing problems involving sparse regularizations.
Durif G., Modolo L., Mold J.E., Lambert-Lacroix S., Picard F., 2018. High dimensional classification with combined adaptive sparse PLS and logistic regression. Bioinformatics 34, 485–493. OUP arXiv HAL
PhD manuscript. Multivariate analysis of high-throughput sequencing data. Lyon University, 2016. HAL
Durif G., Modolo L., Mold J.E., Lambert-Lacroix S., Picard F., April 2018. Probabilistic Count Matrix Factorization for Single Cell Expression Data Analysis. RECOMB 2018, Paris (France).
Durif G., Modolo L., Mold J.E., Lambert-Lacroix S., Picard F., April 2017. Count-based Probabilistic PCA for single-cell data analysis. ISCB NGS'2017 Structural Variation and Population Genomics, Barcelona Biomedical Research Park, Barcelona (Spain).
Durif G., Picard F, Lambert-Lacroix S., June 2016. Factorization of count matrices with application to single cell gene expression profile analysis, Journées Ouvertes de Biologie Informatique et Mathématiques (JOBIM) 2016, ENS de Lyon, Lyon (France).
More can be found on the dedicated page.