The XXIX International Conference on Soft Computing and Measurement (SCM'2026) will take place on May 20 - 22, 2026 at Saint Petersburg Electrotechnical University "LETI", St. Petersburg, Russia.

May, 21, 17:00 – 17:30, Telemost 1
KEYNOTE SPEECH 
WITOLD PEDRYCZ, Honorary Chairman of SCM, prof., Department of Electrical & Computer Engineering University of Alberta, Edmonton, Canada
Informed Machine Learning: A Holistic Data – Knowledge Design Environment
Abstract:
Machine Learning (ML) and Artificial Intelligence (AI) have enjoyed a lot of interest and led to numerous success stories including those in areas of high criticality. With the passage of time, some limitations of the ML technology have become visible and raised concerns about the deployment of the ML constructs (including LLMs) and their exclusive reliance on data. Indeed, data are a lifeblood of design methodologies and drive current commonly encountered development practices. At the center of the ML methodology lies a default assumption that the data fully represent the problem to be solved (e.g., classification or prediction). Enormous masses of data are the blessing and the curse. We look at the problem and produce a solution through the lens of data; in many cases, this may lead to the data blinding effect. We advocate that a holistic knowledge-data development perspective is urgently needed.
An Informed ML (IML) has emerged as a new and promising direction of research addressing these needs. In brief, IML is sought as a methodology in which data and knowledge are used in unison to design ML systems. From the design perspective encountered in the ML learning environment, data and knowledge are radically different. Data are numeric and precise. Knowledge is general and usually expressed at the higher level of abstraction (generality). Knowledge and data emerge at different levels of information granularity.
In this talk, we deliver a comprehensive taxonomy of main pursuits of IML and link them with the main ways the knowledge is represented. A historical perspective is offered by studying the symbolic and subsymbolic processing encountered in successive decades of AI.
The two general categories of physics-oriented and neuro-symbolic constructs associated with the ways in which knowledge and data are explored together. We elaborate on the design process being guided by a prudently augmented additive loss function whose corresponding parts minimize distances between the developed ML model and numeric target values and deliver adherence of the model to information granules reflecting available knowledge. A general taxonomy of neuro-symbolic systems involving learning-for-reasoning, reasoning-for-learning, reasoning-learning is discussed. 

Topics of the Conference

Methods and Systems of Artificial Intelligence and Soft Computing

Measurement Theory. Methods and Tools of Measurement

Applied Artificial Intelligence and Measurement Systems

Sessions

Russian and English are the official languages of the conference. 

The Conference Proceedings will be published digitally and distributed among the participants at the opening of the Conference.
Accepted papers will be submitted for inclusion into IEEE Xplore subject to meeting IEEE Xplore’s scope and quality requirements.