# A Local Approach to Forward Model Learning: Results on the Game of Life Game

## TL;DR

This paper presents a local learning approach for forward models in the Game of Life, demonstrating effective prediction and planning even with imperfect models, using various learning methods.

## Contribution

It introduces a novel local cell transition learning method and compares different models, including neural networks and decision trees, for forward model learning in a game setting.

## Key findings

- Local models can be learned with few data points.
- Imperfect models still enable effective game planning.
- Decision trees and neural networks both perform well in this context.

## Abstract

This paper investigates the effect of learning a forward model on the performance of a statistical forward planning agent. We transform Conway's Game of Life simulation into a single-player game where the objective can be either to preserve as much life as possible or to extinguish all life as quickly as possible. In order to learn the forward model of the game, we formulate the problem in a novel way that learns the local cell transition function by creating a set of supervised training data and predicting the next state of each cell in the grid based on its current state and immediate neighbours. Using this method we are able to harvest sufficient data to learn perfect forward models by observing only a few complete state transitions, using either a look-up table, a decision tree or a neural network. In contrast, learning the complete state transition function is a much harder task and our initial efforts to do this using deep convolutional auto-encoders were less successful. We also investigate the effects of imperfect learned models on prediction errors and game-playing performance, and show that even models with significant errors can provide good performance.