Finding Algebraic Mathematical Models from Experimental Data with Artificial Intelligence

Artificial intelligence tools are nowadays used daily to tackle very complex problems. Generative AI has recently led to, at least, three Nobel prizes. However, these undoubtedly sophisticated and effective tools are all representatives of the “black box” paradigm. Ultimately, we know how to design these systems, but not how the underlying parameters interact with one another at the deepest level. In this book you will find a method to unveil these interactions by applying the proven mathematical known properties of neural networks and evolutionary computation. A comprehensive analysis and description of every step in the technique is made. Case studies are included to exemplify its applicability.

If you ever wondered why the convoluted relations taken advantage of in modern AI tools are rarely deeply understood it is because they are generally not known. Models with explanatory properties were lacking. This book addresses this problem by presenting a self-contained and formally consistent methodology. It will lead you not only to arrive at complete understanding of the models arising from the data. It will also lead you to a method yielding models with explanatory properties: the way the variables interact with one another and how they do so. Various case studies arising from this methodology are included.