Moore, James
TextWorld: A Learning Environment for Text-based Games
Côté, Marc-Alexandre, Kádár, Ákos, Yuan, Xingdi, Kybartas, Ben, Barnes, Tavian, Fine, Emery, Moore, James, Hausknecht, Matthew, Asri, Layla El, Adada, Mahmoud, Tay, Wendy, Trischler, Adam
We introduce TextWorld, a sandbox learning environment for the training and evaluation of RL agents on text-based games. TextWorld is a Python library that handles interactive play-through of text games, as well as backend functions like state tracking and reward assignment. It comes with a curated list of games whose features and challenges we have analyzed. More significantly, it enables users to handcraft or automatically generate new games. Its generative mechanisms give precise control over the difficulty, scope, and language of constructed games, and can be used to relax challenges inherent to commercial text games like partial observability and sparse rewards. By generating sets of varied but similar games, TextWorld can also be used to study generalization and transfer learning. We cast text-based games in the Reinforcement Learning formalism, use our framework to develop a set of benchmark games, and evaluate several baseline agents on this set and the curated list.
Research in Progress at the Information Sciences Institute, University of Southern California
Balzer, Robert, Erman, Lee, Feather, Martin, Goldman, Neil, London, Philip, Wile, David, Wilczynski, David, Lingard, Robert, Mark, William, Mann, William, Moore, James, Pirtle, Mel, Dyer, David, Rizzi, William, Cohen, Danny, Barnett, Jeff, Kameny, Iris, Yemini, Yechiam
Over the past two years we have started a program of On the theoretical side, Professor Randall Davis has research into the development of VLSI systems. They have introduced a descriptive formalism called OMEGA, which contributes to many of the issues of Traditional automated synthesis techniques for circuit current concern in knowlege representation, and they have design are restricted to small classes of circuit functions for applied it to describe the various structured entities such as which mathematical methods exist. Sussman and his group have developed computer-aided design tools that can be of much broader assistance. Guy L. Steele developed a language to support such programming, Johan de Kleer studied causal and Professor Marvin Minsky has worked on a theory of human teleological reasoning in the recognition of circuit function thinking, which likens the mind to a society of agents and from schematics, and Howie Shrobe has worked on constraint attempts to combine a number of insights from satisfaction and the development of an interactive knowledgebased psychoanalytic, developmental, and cognitive theories of system for substantially supporting VLSI design. Further work by Richard Greenblatt and Dr. Lucia Doyle has studied belief revision via truth maintenance and Vaina develops the idea of thread memory.