Gaming the Known and Unknown via Puzzle Solving With an Artificial Intelligence Agent

Researchers design a number of methods for an synthetic clever (AI) agent to unravel a stochastic puzzle like Minesweeper.

For many years, efforts in fixing video games had been unique to fixing two-player video games (i.e., board video games like checkers, chess-like video games, and many others.), the place the sport end result might be accurately and effectively predicted by making use of some synthetic intelligence (AI) search approach and accumulating an enormous quantity of gameplay statistics. Nevertheless, such a way and approach can’t be utilized on to the puzzle-solving area since puzzles are typically performed alone (single-player) and have distinctive traits (comparable to stochastic or hidden info). So then, a query arose as to how the AI approach can retain its efficiency for fixing two-player video games however as a substitute utilized to a single-agent puzzle?

For years, puzzles and video games had been thought to be interchangeable or one a part of the different. In reality, this is probably not the case all the time. Wanting from a real-world perspective, ‘game’ is one thing we face day-after-day; coping with the unknown. For example, the unknown of constructing the proper choice (i.e., getting married) or the improper one (i.e., quitting a job) or not making one in any respect (i.e., regrets on ‘what if’). In the meantime, ‘puzzle’ is one thing that was identified to be there, and even one thing is hidden but to be uncovered. Such a identified case, as an example, can be the discovery of ‘wonder’ materials like graphene and its many potentials which might be but to be commercialized and broadly used. Then once more, how and what border between ‘puzzle’ and ‘game’ in a puzzle-solving context?

At the Japan Superior Institute of Science and Expertise (JAIST), Japan, Professor Hiroyuki Iida, and colleagues tried to reply these two questions of their newest examine printed in the journal Data-based Methods. The analysis examine focuses on two essential contributions: (1) defining the solvability of a puzzle in a single-agent sport context via Minesweeper testbed and (2) proposing a brand new synthetic intelligence (AI) agent utilizing the unified composition of 4 methods known as PAFG solver. Benefiting from the identified info and unknown info of the Minesweeper puzzle, the proposed solver had achieved higher efficiency in fixing the puzzle corresponding to the state-of-the-art research.

AI Boundary of Solvability

The determine depicts AI methods that use knowledge-driven methods to cope with unknown info whereas adopting data-driven methods to make use of the identified info of the Minesweeper puzzle. The resultant findings set up the boundary situation for solvability in a single-player stochastic puzzle which is canonical to broad real-world issues. Credit score: Hiroyuki Iida from JAIST

The researchers adopted an AI agent composed of two knowledge-driven methods and two data-driven methods to finest use the identified and unknown info of the present choice to finest estimate the subsequent choice to make. In consequence, the boundary between the puzzle-solving and game-playing paradigm might be established for the single-agent stochastic puzzle like the Minesweeper.

Such a situation performs a very essential position in real-world issues the place the boundary between the identified and unknown is often blurred and very exhausting to determine. As Professor Iida remarks: “With the capability of AI agent to enhance puzzle solving performance, the boundary of solvability become apparent. Such a situation allowed the clear definition of ‘puzzle’ and ‘game’ conditions, typically found in many real-life situations, such as determining high-stake investment, assessing the risk level of an important decision, and so on.” In essence, all of us reside in our Minesweeper world, making an attempt to guess our means ahead whereas avoiding the ‘bomb’ in our life.

Many uncertainties existed with the face-paced development of present expertise and new paradigm of computing obtainable (i.e., IoT, cloud-based companies, edge computing, neuromorphic computing, and many others.). This situation may very well be true for folks (i.e., technological affordance), neighborhood (i.e., expertise acceptance), society (i.e., tradition and norm), and even at the nationwide ranges (i.e., coverage and guidelines adjustments). “Every day human activity involves a lot of ‘game’ and ‘puzzle’ conditions. However, mapping the solvability paradigm at scale, boundary conditions between the known and the unknown can be established, minimizing the risk of the unknown and maximizing the benefit of the known,” explains Ms. Chang Liu, the lead creator of the examine. “Such a feat is achieved by culminating knowledge-driven techniques, AI technology, and measurable uncertainty (such as winning rate, success rate, progress rate, etc.) while still keeping the puzzle fun and challenging.”

Reference: “A solver of single-agent stochastic puzzle: A case study with Minesweeper” by Chang Liu, Shunqi Huang, Gao Naying, Mohd Nor Akmal Khalid and Hiroyuki Iida, 28 March 2022, Data-Primarily based Methods.
DOI: 10.1016/j.knosys.2022.108630

About Japan Superior Institute of Science and Expertise, Japan

Based in 1990 in Ishikawa prefecture, the Japan Superior Institute of Science and Expertise (JAIST) was the first impartial nationwide graduate college in Japan. Now, after 30 years of regular progress, JAIST has change into certainly one of Japan’s top-ranking universities. JAIST counts with a number of satellite tv for pc campuses and strives to foster succesful leaders with a state-of-the-art schooling system the place range is essential; about 40% of its alumni are worldwide college students. The college has a novel model of graduate schooling primarily based on a fastidiously designed coursework-oriented curriculum to make sure that its college students have a stable basis on which to hold out cutting-edge analysis. JAIST additionally works intently each with native and abroad communities by selling trade–academia collaborative analysis.

About Ms. Chang Liu from Japan Superior Institute of Science and Expertise, Japan

Ms. Chang Liu is a doctoral scholar at the College of Superior Science and Expertise (JAIST), Nomi, Japan. Her analysis focuses on researching attraction details about the evolution of puzzle video games primarily based on the sport mechanics and participant’s expertise, supervised by Professor Hiroyuki Iida in the Lab of Leisure Expertise. She is engaged on analyzing the vital components in the evolution of historic to fashionable puzzle video games, and the info evaluation throughout the means of fixing puzzles and enjoying video games, to discovering a line between puzzles and video games.

About Professor Hiroyuki Iida from Japan Superior Institute of Science and Expertise, Japan

Dr. Hiroyuki Iida obtained his Ph.D. in 1994 on Heuristic Theories on Recreation-Tree Search from the Tokyo College of Agriculture and Expertise, Japan. Since 2005, he has been a Professor at JAIST, the place he’s additionally a Trustee and Vice President of Academic and Pupil Affairs. He’s the head of the Iida laboratory and has printed over 300 papers, displays, and books. His analysis pursuits embrace synthetic intelligence, sport informatics, sport principle, mathematical modeling, search algorithms, game-refinement principle, sport tree search, and leisure science.

Funding info

This examine was funded by a grant from the Japan Society for the Promotion of Science in the framework of the Grant-in-Support for Difficult Exploratory Analysis (Grant Quantity 19K22893).

For wordpress themes and plugins, go to the weblog .

Leave a Comment

Ads Blocker Image Powered by Code Help Pro

Ads Blocker Detected!!!

We have detected that you are using extensions to block ads. Please support us by disabling these ads blocker.