Towards Multi-Agent Communication-Based Language Learning

Abstract

We propose an interactive multimodal framework for language learning. Instead of being passively exposed to large amounts of natural text, our learners (implemented as feed-forward neural networks) engage in cooperative referential games starting from a tabula rasa setup, and thus develop their own language from the need to communicate in order to succeed at the game. Preliminary experiments provide promising results, but also suggest that it is important to ensure that agents trained in this way do not develop an adhoc communication code only effective for the game they are playing

9 pages, manuscript under submission

References in corpus (2)

• Very Deep Convolutional Networks for Large-Scale Image Recognition
• Dialog-based Language Learning

Cited by in corpus (7)

• A Deep Reinforcement Learning Chatbot
• Emergence of Communication in an Interactive World with Consistent Speakers
• Measuring Compositionality in Representation Learning
• A Paradigm for Situated and Goal-Driven Language Learning
• Zero-Resource Neural Machine Translation with Multi-Agent Communication Game
• Document-editing Assistants and Model-based Reinforcement Learning as a Path to Conversational AI
• Analogs of Linguistic Structure in Deep Representations