Can Monte-Carlo Tree Search learn to sacrifice?

Companez, Nathan; Aleti, Aldeida

doi:10.1007/s10732-016-9320-y

Cited by 8 publications

(4 citation statements)

References 33 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…A different, more technical notion of "sacrifice" emerges from Companez and Aleti (2016), which focuses on the Monte-Carlo Tree Search (MCTS) algorithm's ability to identify and execute sacrifice moves in games. In this context, sacrifice moves are defined as strategically losing a resource to gain a long-term advantage.…”

Section: Perspectives On Techno-sacrificementioning

confidence: 99%

Technology and Sacrifice

Leone

2024

Religions

View full text Add to dashboard Cite

This paper investigates the complex relationship between technology, religion, and sacrifice, positing this last term as a pivotal concept for understanding the evolution and impact of technological advancements. Through a detailed examination of various cultural and religious frameworks, it explores how artificial intelligence and other modern technologies both challenge and redefine traditional notions of the sacred and the profane. By analyzing historical and contemporary practices, the study highlights the paradoxical role of sacrifice in the digital age, serving as both a metaphor for the loss inherent in technological progress and a foundational principle that shapes the ethical landscapes of innovation.

show abstract

Section: Perspectives On Techno-sacrificementioning

confidence: 99%

Technology and Sacrifice

Leone

2024

Religions

View full text Add to dashboard Cite

show abstract

“…by introducing score gains which guide the AI player away from winning the game), using agents from the General Video Game AI Competition (GVGAI) to show that AI game players can be easily tricked into not finding the optimal solution. Companez et al (Companez and Aleti 2016) look at enhancements for Monte Carlo Tree Search in Tic-Tac-Toe variations meant to overcome such deceptive issues, highlighting a particular situation where the agent should be able to self-sacrifice in the short run in order to obtain a larger gain in the long run.…”

Section: Literature Reviewmentioning

confidence: 99%

Tackling Sparse Rewards in Real-Time Games with Statistical Forward Planning Methods

Gaina

Lucas

Peŕez-Liebana

2019

AAAI

View full text Add to dashboard Cite

One of the issues general AI game players are required to deal with is the different reward systems in the variety of games they are expected to be able to play at a high level. Some games may present plentiful rewards which the agents can use to guide their search for the best solution, whereas others feature sparse reward landscapes that provide little information to the agents. The work presented in this paper focuses on the latter case, which most agents struggle with. Thus, modifications are proposed for two algorithms, Monte Carlo Tree Search and Rolling Horizon Evolutionary Algorithms, aiming at improving performance in this type of games while maintaining overall win rate across those where rewards are plentiful. Results show that longer rollouts and individual lengths, either fixed or responsive to changes in fitness landscape features, lead to a boost of performance in the games during testing without being detrimental to non-sparse reward scenarios.

show abstract

“…Artificial Intelligence (AI) is the generic name given to the development of programs attempting to construct an artificial form of human intelligence. AIs allow problems to be solved through learning and through inference of information with associative memory and systems based on knowledge (Çaliş & Bulkan, 2015;Companez & Aleti, 2016). The search methods consist of a set of elements, each being identified by a special key.…”

Section: Introductionmentioning

confidence: 99%

Contributions to the future of metaheuristics in the contours of scientific development

et al. 2022

View full text Add to dashboard Cite

Metaheuristic algorithms solve optimisation problems by identifying the best combination among a set of variables to enhance a function. Within metaheuristics, the main purpose of this work is that of showing the development of research issues about processes related to optimisation and metaheuristics, with a focus on the projection of those issues with greater possibility of development. Optimization processes is one of the most studied fields in artificial intelligence, optimization, logistics, and other applications The main contributions of this work were the identification of the main issues contained in the themes of process optimization and metaheuristics; an analysis of the expansion and retraction of the aforementioned theme; an understanding of convergence and divergence; and an analysis of the stages of development as presented in the gaps of the fifty most commonly mentioned articles. The main finding was to analyze the development of research topics on optimization processes and metaheuristics, focusing on projecting the topics most likely to develop.

show abstract

Can Monte-Carlo Tree Search learn to sacrifice?

Cited by 8 publications

References 33 publications

Technology and Sacrifice

Technology and Sacrifice

Tackling Sparse Rewards in Real-Time Games with Statistical Forward Planning Methods

Contributions to the future of metaheuristics in the contours of scientific development

Contact Info

Product

Resources

About