Researchers design several strategies for a synthetic intelligent (AI) representative to fix a stochastic puzzle like Minesweeper.
For decades, efforts in resolving video games had been exclusive to resolving two-player video games (i.e., parlor game like checkers, chess-like games, and so on), where the game result can be properly and efficiently forecasted by using some artificial intelligence (AI) search technique and collecting an enormous quantity of gameplay data. However, such an approach and method can not be applied straight to the puzzle-solving domain since puzzles are typically played alone (single-player) and have distinct qualities (such as concealed or stochastic details). Then, a concern arose as to how the AI strategy can keep its performance for resolving two-player games however rather used to a single-agent puzzle?
Then, a question arose as to how the AI strategy can maintain its efficiency for resolving two-player video games but instead used to a single-agent puzzle?
For games, years and puzzles had actually been concerned as interchangeable or one part of the other. puzzle is something that was understood to be there, and even something is concealed yet to be uncovered. Once again, how and what border in between puzzle and video game in a puzzle-solving context?
At the Japan Advanced Institute of Science and Technology (JAIST), Japan, Professor Hiroyuki Iida, and colleagues tried to respond to these two questions in their most current study released in the journal Knowledge-based Systems. The research study focuses on two crucial contributions: (1) defining the solvability of a puzzle in a single-agent game context through Minesweeper testbed and (2) proposing a new expert system (AI) agent using the combined composition of 4 methods called PAFG solver. Making the most of the understood info and unidentified information of the Minesweeper puzzle, the proposed solver had attained much better performance in resolving the puzzle equivalent to the modern studies.
The figure portrays AI techniques that utilize knowledge-driven techniques to handle unknown info while adopting data-driven techniques to utilize the understood details of the Minesweeper puzzle. The resultant findings develop the boundary condition for solvability in a single-player stochastic puzzle which is canonical to broad real-world issues. Credit: Hiroyuki Iida from JAIST
The scientists adopted an AI representative composed of two knowledge-driven techniques and two data-driven methods to finest utilize the recognized and unidentified details of the existing decision to best estimate the subsequent choice to make. As a result, the boundary between the puzzle-solving and game-playing paradigm can be developed for the single-agent stochastic puzzle like the Minesweeper.
As Professor Iida remarks: “With the ability of AI representative to enhance puzzle resolving performance, the boundary of solvability end up being evident. In essence, we all live in our Minesweeper world, trying to guess our method forward while avoiding the bomb in our life.
Lots of unpredictabilities existed with the face-paced development of existing innovation and new paradigm of computing offered (i.e., IoT, cloud-based services, edge computing, neuromorphic computing, and so on). This condition could be true for individuals (i.e., technological affordance), community (i.e., technology acceptance), society (i.e., culture and standard), and even at the national levels (i.e., policy and guidelines changes). “Every day human activity includes a great deal of game and puzzle conditions. Nevertheless, mapping the solvability paradigm at scale, boundary conditions in between the known and the unknown can be developed, decreasing the danger of the unknown and taking full advantage of the advantage of the known,” discusses Ms. Chang Liu, the lead author of the study. “Such a feat is accomplished by culminating knowledge-driven techniques, AI technology, and measurable uncertainty (such as winning rate, success rate, development rate, etc) while still keeping the puzzle enjoyable and challenging.”
Recommendation: “A solver of single-agent stochastic puzzle: A case research study with Minesweeper” by Chang Liu, Shunqi Huang, Gao Naying, Mohd Nor Akmal Khalid and Hiroyuki Iida, 28 March 2022, Knowledge-Based Systems.DOI: 10.1016/ j.knosys.2022.108630.
About Japan Advanced Institute of Science and Technology, Japan.
Established in 1990 in Ishikawa prefecture, the Japan Advanced Institute of Science and Technology (JAIST) was the first independent nationwide graduate school in Japan. Now, after 30 years of steady development, JAIST has actually turned into one of Japans top-ranking universities. JAIST counts with numerous satellite campuses and aims to cultivate capable leaders with a cutting edge education system where variety is key; about 40% of its alumni are worldwide students. The university has a special style of graduate education based upon a carefully designed coursework-oriented curriculum to guarantee that its trainees have a strong structure on which to perform cutting-edge research study. JAIST also works closely both with regional and overseas neighborhoods by promoting industry– academia collaborative research.
About Ms. Chang Liu from Japan Advanced Institute of Science and Technology, Japan.
Ms. Chang Liu is a doctoral trainee at the School of Advanced Science and Technology (JAIST), Nomi, Japan. Her research study focuses on researching appeal info about the evolution of puzzle games based upon the game mechanics and gamers experience, supervised by Professor Hiroyuki Iida in the Lab of Entertainment Technology. She is working on analyzing the significant factors in the development of ancient to modern puzzle games, and the info analysis during the procedure of fixing puzzles and playing games, to finding a line in between video games and puzzles.
About Professor Hiroyuki Iida from Japan Advanced Institute of Science and Technology, Japan.
Dr. Hiroyuki Iida received his Ph.D. in 1994 on Heuristic Theories on Game-Tree Search from the Tokyo University of Agriculture and Technology, Japan. His research study interests include artificial intelligence, game informatics, video game theory, mathematical modeling, search algorithms, game-refinement theory, game tree search, and entertainment science.
Financing details.
This study was funded by a grant from the Japan Society for the Promotion of Science in the framework of the Grant-in-Aid for Challenging Exploratory Research (Grant Number 19K22893).
The research study focuses on 2 essential contributions: (1) specifying the solvability of a puzzle in a single-agent video game context by means of Minesweeper testbed and (2) proposing a new synthetic intelligence (AI) agent utilizing the merged composition of 4 techniques called PAFG solver. Taking advantage of the understood details and unidentified details of the Minesweeper puzzle, the proposed solver had actually accomplished better efficiency in resolving the puzzle similar to the cutting edge research studies.
Her research focuses on looking into appeal details about the development of puzzle video games based on the game mechanics and gamers experience, monitored by Professor Hiroyuki Iida in the Lab of Entertainment Technology. She is working on examining the substantial factors in the evolution of ancient to contemporary puzzle games, and the information analysis throughout the procedure of resolving puzzles and playing video games, to finding a line in between puzzles and games.