Information-Induced Geometries in Statistics, Data Analysis, and Applied Research

Dear Colleagues,

Many statistical and data analysis procedures, including least squares estimation and hierarchical classification methods, rely on geometric techniques, whether implicitly or explicitly. These techniques are essential in representing statistical populations or individual data points within low-dimensional Euclidean spaces. Choosing the appropriate geometric framework for data analysis is crucial, as it directly impacts the effectiveness and properties of the statistical methods developed.

From a methodological standpoint, it is beneficial to center our attention on the formal characteristics of the observation process. These characteristics, which encompass various dimensions of how we gather and interpret data, can be precisely articulated through the lens of information theory. By achieving this, we can make informed decisions regarding the most suitable geometric framework, ultimately enriching our analytical approach and enhancing our insights into the observed phenomena.

This naturally leads us to consider the informative geometry, such as the natural geometry in data analysis and statistics. This geometry is the Riemannian geometry induced in the parameter space of a parametric statistical model through the well-known Fisher information matrix. It allows us to obtain many other statistical distances in a unified form. The introduced Riemannian geometry is not only valuable in data analysis as a measure of similarity between different statistical populations and a proper generalization of the well-known Mahalanobis distance, it can also be used in fundamental aspects of statistics, such as point estimation, as it allows us to define intrinsic versions of the bias and the quadratic error independently of the model parametrization. 

Understanding complex data structures and relationships through geometric interpretation is fundamental in the dynamic world of statistics and data analysis. This approach helps us enhance our knowledge and improve our ability to make informed data-driven decisions, especially with machine learning and artificial intelligence, which has transformed the exploration of information-induced geometries. As these technologies rely heavily on extracting patterns from large datasets, the geometric framework provides a powerful tool to understand the behavior of algorithms, optimize performance, and ensure interpretability. 

Information-induced geometries can provide a revolutionary framework, transforming the fields of statistics, data analysis, and applied research. By creating a rich tapestry of relationships and structures, this approach allows researchers and analysts to uncover patterns and insights within data that were previously obscured. In particular, physics stands to gain tremendously from this methodology, as it can facilitate a deeper exploration of the fundamental principles governing the universe, ultimately leading to discoveries and advancements in this field.

This Special Issue will showcase the multifaceted nature and vital importance of information-induced geometries. These geometries provide researchers with a range of flexible and robust mathematical tools, essential in thoroughly analyzing complex datasets. We welcome contributions that present recently developed statistical methods, illustrating their practical applicability and effectiveness in addressing real-world challenges across various fields by analyzing diverse and intricate data. 

Prof. Dr. Josep Maria Oller
Dr. David Bernal-Casas
Guest Editors

Manuscript Submission Information

Manuscripts should be submitted online at www.mdpi.com by registering and logging in to this website. Once you are registered, click here to go to the submission form. Manuscripts can be submitted until the deadline. All submissions that pass pre-check are peer-reviewed. Accepted papers will be published continuously in the journal (as soon as accepted) and will be listed together on the special issue website. Research articles, review articles as well as short communications are invited. For planned papers, a title and short abstract (about 250 words) can be sent to the Editorial Office for assessment.

Submitted manuscripts should not have been published previously, nor be under consideration for publication elsewhere (except conference proceedings papers). All manuscripts are thoroughly refereed through a single-anonymized peer-review process. A guide for authors and other relevant information for submission of manuscripts is available on the Instructions for Authors page. Mathematics is an international peer-reviewed open access semimonthly journal published by MDPI.

Please visit the Instructions for Authors page before submitting a manuscript. The Article Processing Charge (APC) for publication in this open access journal is 2600 CHF (Swiss Francs). Submitted papers should be well formatted and use good English. Authors may use MDPI's English editing service prior to publication or during author revisions.

• information
• positive-definite kernel
• information geometry
• intrinsic statistical data analysis
• principle of minimum loss of Fisher information
• variational principles
• quantum mechanics
• Bayes theorem