Dr. Eng. Agus Naba, S.Si., M.T was inaugurated as a professor in the field of Intelligent Systems, Saturday (13/08/2022), in the Samantha Krida building, Universitas Brawijaya (UB). He is the 27th active professor from the Faculty of Mathematics and Natural Sciences, and the 170th active professor at UB, and the 298th of all professors that UB has produced.
In his speech, he offered the “Heuristic Artificial Intelligence Modeling (HAIM)” approach to support industry 4.0.
He explained, intelligent system technology is one of the key technologies and plays a central role as a driving force in industry 4.0. Intelligent systems technology offers a variety of intelligent system modeling techniques to build data-driven intelligent system models. However, there are still some obstacles in the modeling of intelligent systems.
One of the obstacles is that the data available in the industry are generally isolated or specific, inconsistent and of low quality. With poor and complex data quality, processed with standard preprocessing techniques, as well as speculative parameter optimization, it is certainly difficult to expect intelligent system modeling to produce optimal models. The resulting intelligent system model is black-box whose validity is difficult to fully guarantee.
In general, the constraints faced in modeling intelligent systems based on the following three things: 1) modeling is purely data-based (inductive) regardless of knowledge of system behavior, 2) optimization techniques for model parameters are speculative so that they are trapped in local optimum parameters, and 3. ) data preprocessing methods are general in nature without regard to the complexity and uniqueness or characteristics of the data.
According to Agus Naba, the deductive approach is a wise choice as long as system knowledge is available. Optimization techniques need to be made more systematic and focused, no longer speculative. Meanwhile, depending on the case, preprocessing techniques need to be built specifically based on the uniqueness of the system. If these steps can be carried out, the intelligent system modeling will be more effective and efficient and the potential defects of the intelligent system model can be reduced.
To overcome these obstacles, Agus Naba proposed a strategy called Heuristic AI Modeling (HAIM) or heuristic intelligent system modeling. HAIM suggests three things: 1) deductive or combined deductive-inductive approaches are prioritized in intelligent system modeling, 2) speculative optimization algorithms are replaced with more systematic and targeted ones, and 3) preprocessor units for extracting unique data features need to be designed specifically for each case.
Several researches have been successfully solved by Agus Naba by applying the HAIM approach. For example, estimation of wind speed in wind turbines uses a fuzzy model built by heuristic-deductive, optimization of fuzzy model controllers to solve the inverted pendulum problem, detection of pneumonia in X-ray images, detection of car objects, and detection of types of disturbances in the power grid. [Irene/ Humas UB/Trans. Iir]