ARTICLE | doi:10.20944/preprints201905.0090.v1
Subject: Mathematics & Computer Science, Artificial Intelligence & Robotics Keywords: intelligence; inductive methods; deductive methods; pseudorandom number; artificial intelligence; Prolog; Otter; Z3; deep learning; ensemble methods; automated reasoning; coin-weighing puzzles
Online: 8 May 2019 (10:03:46 CEST)
This paper briefly reviews the state of the art in artificial intelligence including inductive and deductive methods. Deep learning and ensemble machine learning lie in inductive methods while automated reasoning implemented in deductive computer languages (Prolog, Otter, and Z3) is based on deductive methods. In the inductive methods, intelligence is inferred by pseudorandom number for creating the sophisticated decision trees in Go (game), Shogi (game), and quiz bowl questions. This paper demonstrates how to wisely use the pseudorandom number for solving coin-weighing puzzles with the deductive method. Monte Carlo approach is a general purpose problem-solving method using random number. The proposed method using pseudorandom number lies in one of Monte Carlo methods. In the proposed method, pseudorandom number plays a key role in generating constrained solution candidates for coin-weighing puzzles. This may be the first attempt that every solution candidate is solely generated by pseudorandom number while deductive rules are used for verifying solution candidates. In this paper, the performance of the proposed method was measured by comparing with the existing open source codes by solving 12-coin and 24-coin puzzles respectively.