Arnaud Mazier: Decision Trees methods, an overview of the white-boxes

Machine Learning Seminar presentation

Topic: Data-Driven Hyper-elastic Simulations

Speaker: Arnaud Mazier, University of Luxembourg, Department of Computational Science

Time: Wednesday 2020.11.04, 10:00 CET

How to join: Please contact Jakub Lengiewicz

Abstract:

Machine learning methods such as neural networks start to critically impact the medical field due to fast and reliable algorithms. The main drawbacks of these methods are the enormous quantity of data needed, a long training phase, and a black-box algorithm.

In this talk, I will give a brief overview of Decision Trees (DT) models. DTs are a supervised learning algorithm that predicts the values of a target variable by learning simple decision rules from the data. These algorithms are considered as white-boxes as they can expose the decisions made and are fast to train. Random Forests (RF) and Extremely Randomized Trees (ERTs) are tree-based ensemble methods, i.e., they combine several trees for improving results over a single estimator; they are considered state-of-the-art methods in machine learning.

Additional material:

[1] Martínez-Martínez et al. A finite element-based machine learning approach for modeling the mechanical behavior of the breast tissues under compression in real-time. Computers in Biology and Medicine. https://doi.org/10.1016/j.compbiomed.2017.09.019

[2] Andrea Mendizabal, Eleonora Tagliabue, Jean-Nicolas Brunet, Diego Dall’Alba, Paolo Fiorini, et al. Physics-based Deep Neural Network for Real-Time Lesion Tracking in Ultrasound-guided Breast Biopsy. https://hal.archives-ouvertes.fr/hal-02311277/