Comprehensible Context-driven Text Game Playing
This addresses the problem of inefficient training for text-based game playing agents, though it is incremental as it builds on existing DQN and CNN methods.
The paper tackles the slow training of text-based game agents by using a fast CNN to encode syntactic context from observed texts and augmenting the reward signal, resulting in a speedup of one order of magnitude and learning a superior agent.
In order to train a computer agent to play a text-based computer game, we must represent each hidden state of the game. A Long Short-Term Memory (LSTM) model running over observed texts is a common choice for state construction. However, a normal Deep Q-learning Network (DQN) for such an agent requires millions of steps of training or more to converge. As such, an LSTM-based DQN can take tens of days to finish the training process. Though we can use a Convolutional Neural Network (CNN) as a text-encoder to construct states much faster than the LSTM, doing so without an understanding of the syntactic context of the words being analyzed can slow convergence. In this paper, we use a fast CNN to encode position- and syntax-oriented structures extracted from observed texts as states. We additionally augment the reward signal in a universal and practical manner. Together, we show that our improvements can not only speed up the process by one order of magnitude but also learn a superior agent.