Compositional Obverter Communication Learning From Raw Visual Input

Abstract

One of the distinguishing aspects of human language is its compositionality, which allows us to describe complex environments with limited vocabulary. Previously, it has been shown that neural network agents can learn to communicate in a highly structured, possibly compositional language based on disentangled input (e.g. hand- engineered features). Humans, however, do not learn to communicate based on well-summarized features. In this work, we train neural agents to simultaneously develop visual perception from raw image pixels, and learn to communicate with a sequence of discrete symbols. The agents play an image description game where the image contains factors such as colors and shapes. We train the agents using the obverter technique where an agent introspects to generate messages that maximize its own understanding. Through qualitative analysis, visualization and a zero-shot test, we show that the agents can develop, out of raw image pixels, a language with compositional properties, given a proper pressure from the environment.

arXiv commentPublished as a conference paper at ICLR 2018

References in corpus (4)

• Learning to Communicate with Deep Multi-Agent Reinforcement Learning
• Emergence of Grounded Compositional Language in Multi-Agent Populations
• Multi-Agent Cooperation and the Emergence of (Natural) Language
• Learning to Play Guess Who? and Inventing a Grounded Language as a Consequence

Cited by in corpus (11)

• Emergence of Compositional Language with Deep Generational Transmission
• Emergence of Communication in an Interactive World with Consistent Speakers
• Learning by Abstraction: The Neural State Machine
• Towards More Human-like AI Communication: A Review of Emergent Communication Research
• Emergence of Pragmatics from Referential Game between Theory of Mind Agents
• Emergent Communication of Generalizations
• Incorporating Pragmatic Reasoning Communication into Emergent Language
• Entropy Minimization In Emergent Languages
• Catalytic Role Of Noise And Necessity Of Inductive Biases In The Emergence Of Compositional Communication
• Generalizing Emergent Communication
• On Memory Mechanism in Multi-Agent Reinforcement Learning