a. 1980s: Knowledge acquisition eventually became automated, although it did not unfold in the manner initially envisioned. The shift starts from rule-based systems to algorithms that can learn from data, inspired by the concept of neural networks from the 1950s. ”Explanation Based Learning (EBL)” is introduced, a concept wherein a computer analyzes training data to formulate general rules, enabling the elimination of less essential data (DeJong & Mooney, 1986). NetTalk, a system that learns word pronunciation in a manner resembling a child’s natural learning process, is also introduced (Sejnowski & Rosenberg, 1988).
b. 1990s: Decision trees, support vector machines, and other ML algorithms become prominent. A pivotal element in AI is the decision tree, an influential and adaptable instrument that has been instrumental in advancing the realms of machine learning and data mining methodologies. Although decision trees can be traced to the 1950s and 1960, in the 1990s ensemble methods came to the fore, uniting numerous decision trees to enhance the overall performance and precision of models (Frąckiewicz, 2023). The support vector machine (SVM), developed in 1963 and further refined by Vapnik in the 1990s, is another classification tool. These are powerful yet flexible supervised machine learning algorithms that fundamentally represent various classes within a multidimensional space through a hyperplane. SVM iteratively generates this hyperplane to minimize errors. The primary objective of SVM is to partition datasets into classes while seeking a maximum margin hyperplane (MMH) (Bandgar, 2021).