RINA - Démonstrateur d’utilisation de l’intelligence artificielle pour la gestion opérationnelle des risques naturels gravitaires géologiques | FEREC 2020

Figure 2: Passive protective structures at the base of a bedrock. (a) Embankment made of gabions (b) ditch caved and embankment raised with material of the ditch. 1.2. Data analysis Data analysis is observing a dataset and using it to understand the behaviour of a cliff. It helps to find patterns and trends, which may be used to predict future events such as rock falls. This study will be using Artificial Intelligence (AI) for predictions. AI is a section of computer sciences that attempts to create machines capable of performing tasks that require human intelligence. This term is evolving with time and scientists aim to create more and more complex systems. One of the many challenges is providing machines with the capabilities of having creativity and producing things like screen plays. One aspect that has been deeply explored is pattern recognition. The capability of looking at an information, receiving new information and fitting this new information inside what is already known. Many different types of models exist in order to do this. Three of those methods, nearest neighbors, decision tree and neural networks (NNs), will be explored in this work. The branch of machine learning involves pattern recognition, but it differentiates itself by including the concept of learning. It works with the idea of making predictions, comparing their results with the actual values and improving upon them. This is an interactive model. An example of machine learning are NNs and some types of decision trees which will be elaborated on section 3.1.3. 1.3. Aim of the work As technology advances statistical methods grow in performance, speed and applicability. Classical methods have already been used many times to better understand rock falls and the conditions which trigger it. This study aims to: 1. Better understand the causes of rock falls. By creating predictive models based on past events it is desired to better understand which are the most important factors and at what intensities the blocks will fall. 7