Название: Artificial Intelligence for Scientific Discoveries: Extracting Physical Concepts from Experimental Data Using Deep Learning
Автор: Raban Iten
Издательство: Springer
Год: 2023
Страниц: 168
Язык: английский
Формат: pdf (true), epub
Размер: 25.0 MB

Will research soon be done by Artificial Intelligence (AI), thereby making human researchers superfluous? This book explains modern approaches to discovering physical concepts with Machine Learning (ML) and elucidates their strengths and limitations. The automation of the creation of experimental setups and physical models, as well as model testing are discussed. The focus of the book is the automation of an important step of the model creation, namely finding a minimal number of natural parameters that contain sufficient information to make predictions about the considered system. The basic idea of this approach is to employ a Deep Learning (DL) architecture, SciNet, to model a simplified version of a physicist's reasoning process. SciNet finds the relevant physical parameters, like the mass of a particle, from experimental data and makes predictions based on the parameters found. The author demonstrates how to extract conceptual information from such parameters, e.g., Copernicus' conclusion that the solar system is heliocentric.

There has been amazing progress in the field of Artificial Intelligence (AI) and Machine Learning in recent years. Not long ago, robots could not walk, now they can even dance. Computer Vision and Natural Language Processing (NLP) have reached a level of performance that make them indispensable in many areas today, such as face recognition, speech interpretation and self-driving cars. While most AI is developed based on domain knowledge, i.e., specific knowledge about the data under consideration or the environment of interest, there are also impressive results where AI learns from scratch.

As scientists, we are particularly interested in understanding how the Machine Learning system makes its predictions and hence what the underlying model looks like. In this book we focus on this question, i.e., on how we can extract conceptual information from physical systems using Machine Learning. This question is of interest for several reasons. On one hand, it is an important step for the long term goal of building AI physicists, on the other hand, it may help us to gain insight into fundamental problems in modern physics.

This book is written for scientists interested in using Machine Learning to discover physical concepts. No prior knowledge of Machine Learning is required, but a first degree in a science subject is assumed.

Contents: