{
  "id": "1901.09895",
  "version": "v1",
  "published": "2019-01-27T05:06:30.000Z",
  "updated": "2019-01-27T05:06:30.000Z",
  "title": "Modularization of End-to-End Learning: Case Study in Arcade Games",
  "authors": [
    "Andrew Melnik",
    "Sascha Fleer",
    "Malte Schilling",
    "Helge Ritter"
  ],
  "categories": [
    "cs.LG",
    "cs.AI",
    "stat.ML"
  ],
  "abstract": "Complex environments and tasks pose a difficult problem for holistic end-to-end learning approaches. Decomposition of an environment into interacting controllable and non-controllable objects allows supervised learning for non-controllable objects and universal value function approximator learning for controllable objects. Such decomposition should lead to a shorter learning time and better generalisation capability. Here, we consider arcade-game environments as sets of interacting objects (controllable, non-controllable) and propose a set of functional modules that are specialized on mastering different types of interactions in a broad range of environments. The modules utilize regression, supervised learning, and reinforcement learning algorithms. Results of this case study in different Atari games suggest that human-level performance can be achieved by a learning agent within a human amount of game experience (10-15 minutes game time) when a proper decomposition of an environment or a task is provided. However, automatization of such decomposition remains a challenging problem. This case study shows how a model of a causal structure underlying an environment or a task can benefit learning time and generalization capability of the agent, and argues in favor of exploiting modular structure in contrast to using pure end-to-end learning approaches.",
  "revisions": [
    {
      "version": "v1",
      "updated": "2019-01-27T05:06:30.000Z"
    }
  ],
  "analyses": {
    "keywords": [
      "case study",
      "arcade games",
      "value function approximator learning",
      "environment",
      "end-to-end learning approaches"
    ],
    "note": {
      "typesetting": "TeX",
      "pages": 0,
      "language": "en",
      "license": "arXiv",
      "status": "editable"
    }
  }
}