Communications of the ACM
C4.5: programs for machine learning
C4.5: programs for machine learning
| Hi-index | 0.00 |
One writes examples of statements so that one example solves an instance of a problem, and writes how each example runs on a computer. For instance, the one writes a statement "From 1 to 10, repeat the following", and also writes how the statement repeats its execution on the computer. Then the one gives them to an initial program, pI, and lets the initial program generalize how the examples of the statements run and generate a program, pg. When the initial program, pI, gets a new example of statements to solve a new instance of the problem, the program, pI, executes the program, pg. For instance, the one writes a statement "From 5 to 8, repeat the following", and then the program, pG, repeats the statement four times. As a result of generating a program, pG, the initial program, pI, acquires implicitly rules of a grammar that produce statements, and semantics of the statements. Since the generated program, pG, describes how to execute varying number of repetitions, and varying depth of recursive calls, we argue our initial program, pI, acquires a grammar of a language that is equivalent to that used in a conventional programming language.