Analyzing complex and high-dimensional data with deep learning
Seminar held by Frank Emmert-Streib (Predictive Society and Data Analytics Lab, Tampere University) on February 27
27/02/2020
:
10h00
MSI, 1361 Route des Lucioles, 06560 Valbonne, France (InPhyNi and Maison de l'IA Building)
Speaker: Frank Emmert-Streib (Predictive Society and Data Analytics Lab, Tampere University)
Publication :
27/02/2020
Partager cet article :
Speaker: Frank Emmert-Streib (Predictive Society and Data Analytics Lab, Tampere University)
Abstract: In recent years, deep learning gained considerable attention as a new machine learning method throughout the sciences. Deep learning models have been successfully used in many application areas, most notably in image recognition and speech recognition. However, in computational biology, deep learning is still at an early stage and the studied data come mostly from the DNA-level. In contrast, in this talk I present results about the applications of deep learning to gene expression data and electronic health records (eHR) data. In both cases, the classification of patient’s samples from complex disorders, including lung cancer, obesity, chronic pain and psychiatric disorders, is investigated. Overall, our results reveal interesting insights about the information carried by non-coding RNAs and feature selection mechanism controlling the dimensionality of the input.