Knowledge-based expert systems 1960s-1970s
During this period, researchers began to develop systems that simulated the knowledge and reasoning processes of human experts. These systems mainly rely on knowledge provided by domain experts and use rule-based reasoning engines to solve specific problems. For example, Edward Shortliffe developed MYCIN in 1974 for medical diagnosis; Bruce lithuania mobile database Buchanan and Joshua Lederberg developed DENDRAL in 1965 for chemical structure inference. Although these systems have been successful in specific fields, they are difficult to cope with new problems and unknown fields because they rely on manually encoded knowledge. In addition, knowledge-based expert systems require a lot of maintenance costs because the knowledge base needs to be constantly updated. Important papers on MYCIN include Edward Shortliffe's "MYCIN: A Rule-Based Computer Program for Advising Physicians Regarding Antimicrobial Therapy Selection" 1974; important papers on DENDRAL include Buchanan and Lederberg et al. "Heuristic DENDRAL: A Program for Generating Explanatory Hypotheses in Organic Chemistry" 1971.
Genetic Algorithms 1960s
In order to find a method to solve complex optimization problems, John Holland drew inspiration from the process of natural selection and inheritance and proposed the genetic algorithm. This is a global optimization method that is widely used in combinatorial optimization problems. The genetic algorithm continuously evolves to solve problems by simulating the process of gene crossover, mutation and selection. However, the genetic algorithm has the limitations of slow convergence, sensitive parameter settings and easy to fall into local optimality. In 1975, John Holland published a book titled "Adaption in Natural and Artificial Systems", which detailed the principles and implementation methods of genetic algorithms.