Dialog-based Language Learning
This work addresses the challenge of more human-like language learning for AI dialog systems, though it appears incremental as it builds on existing datasets and baseline strategies.
The paper tackles the problem of building intelligent dialog agents by proposing dialog-based language learning, where supervision is derived from a dialog partner's responses during conversation, and demonstrates that a novel model with predictive lookahead can learn to answer questions correctly without reward-based supervision.
A long-term goal of machine learning research is to build an intelligent dialog agent. Most research in natural language understanding has focused on learning from fixed training sets of labeled data, with supervision either at the word level (tagging, parsing tasks) or sentence level (question answering, machine translation). This kind of supervision is not realistic of how humans learn, where language is both learned by, and used for, communication. In this work, we study dialog-based language learning, where supervision is given naturally and implicitly in the response of the dialog partner during the conversation. We study this setup in two domains: the bAbI dataset of (Weston et al., 2015) and large-scale question answering from (Dodge et al., 2015). We evaluate a set of baseline learning strategies on these tasks, and show that a novel model incorporating predictive lookahead is a promising approach for learning from a teacher's response. In particular, a surprising result is that it can learn to answer questions correctly without any reward-based supervision at all.