Dr. Sadegh Abbaszadeh

I received my MSc and PhD in mathematics (nonlinear analysis) at Semnan University (Semnan, Iran). After one-year collaboration with “Intelligent System and Perception Recognition Labarotry” at Shahid Beheshti University (Tehran, Iran), I achieved a one-year research fund from “Iran National Science Foundation” for a project in “Nonlinear Integrals and their applications in Decision Making”. Since July 2018, I have been a research associate in machine learning at Intelligent Systems and Machine Learning group of Paderborn University.

Briefly, my research interest is around learning of predictive models that guarantee monotonicity in the input variable, which has received increasing attention in machine learning in recent years. By trend, the difficulty of ensuring monotonicity increases with the flexibility or, say, nonlinearity of a model.

One of the main technical concerns in any learning machine which coordinately affects the output of that machine, is how to properly use an aggregation as a global utility function, in order to combine several values and decide about the right position of a query input. Correct selection of such a function is very important, especially when the aggregating values are not independent of each other. In decision making, this problems is known as a Multi-Criteria
Decision Making. Examples of such combinations in machine learning include methods for learning the majority rule model, the non-compensatory sorting model, decision rule learning and the Choquet integral. In contrast to many other machine learning approaches, corresponding models are interpretable and meaningful from of decision making point of view. Besides, they often guarantee other properties that might be desirable, such as monotonicity.