In this work, we propose a Monte Carlo Tree Search (MCTS) based approach to procedurally generate Sokoban puzzles. Our method generates puzzles through simulated game play, guaranteeing solvability in all generated puzzles. We perform a user study to infer features that are efficient to compute and are highly correlated with expected puzzle difficulty. We combine several of these features into ...

Current domain-independent, classical planners require symbolic models of the problem domain and instance as input, resulting in a knowledge acquisition bottleneck. Meanwhile, although deep learning has achieved significant success many fields, is encoded subsymbolic representation which incompatible with systems such planners. We propose Latplan, an unsupervised architecture combining planning...

There is a fundamental connection between the notions of game and of computation. At its most basic level, this is implied by any game complexity result, but the connection is deeper than this. One example is the concept of alternating nondeterminism, which is intimately connected with two-player games. In the first half of this thesis, I develop the idea of game as computation to a greater deg...

The use of a lower bound estimate in the search has a tremendous impact on the size of the resulting search trees, whereas OBDDs can be used to e ciently describe sets of states based on their binary encoding. This paper combines these two ideas to a new algorithm BDDA . It challenges both, the breadthrst search using OBDDs and the traditional A algorithm. The problem with A is that in many app...

In many problems in commonsense reasoning and intelligent manufacturing, we need to reason about cutting, joining, and folding sheets of materials such as cardboard and metal. In this paper we introduce a first-order ontology of shape (called BoxWorld) that can support these applications. We reuse an existing ontology of shape for object recognition with 2D shapes (called CardWorld) and extend ...

Simple behaviours form the underpinnings of complex high-level behaviours. Traditional AI research focuses on high-level behaviours and ignores the underpinnings that are considered to be irrelevant for the problem at hand. As a rst step towards a hybrid approach to complex behaviour, this paper studies the simple behaviour of box-pushing that underlies complex planning in, for instance, the ga...

A level generator is a tool that generates game levels from noise. Training without dataset suffers feedback sparsity, since it unlikely to generate playable via random exploration. common solution shaped rewards, which guides the achieve subgoals towards playability, but they consume effort design and require game-specific domain knowledge. This paper proposes novel approach train generators d...

The extension of reinforcement learning (RL) to large state space has inevitably encountered the problem of the curse of dimensionality. Improving the learning efficiency of the agent is much more important to the practical application of RL. Consider learning to optimally solve Markov decision problems in a particular domain, if the domain has particular characteristics that are attributable t...

We consider the problem of learning for planning, where knowledge acquired while planning is reused to plan faster in new problem instances. For robotic tasks, among others, plan execution can be captured as a sequence of visual images. For such domains, we propose to use deep neural networks in learning for planning, based on learning a reactive policy that imitates execution traces produced b...