Machine Learning from Conversation with Humans

Human social learning is an effective process that has inspired many existing machine learning techniques, such as learning from observation and learning by demonstration. Hence, in this paper, we are proposing another form of social learning, Learning from a Conversation (LfC). LfC is an open-ended machine learning system in which an artificially intelligent agent learns from extended dialog with a human. Our system enables the agent to adapt to new changes based on the human input. We provide a detailed description of our system and report its performance by providing several examples that reflect our system's efficiency. Test results indicate that the prototype was successful in learning from conversation.