Development of intrinsic motivation complex for an artificial conversational entity

One of the most challenging problems in Machine Learning and AI is the conversation between human and machine. The humanization of the interaction between a user and a machine represents a hot topic in the current research.

On the business side we are looking at a rapidly growing demand for algorithmic personal assistants that are in need of customizing for a large variety of domains. In order to be able to produce human-like interaction and provide reliable assistance to people the key areas of improvement are:

1. Understanding of human intention based on semantic analysis of textual input
2. Optimizing the machine to provide the best response/question in terms of maximal information gain
3. Efficiently generating textual and audio output to serve the human counterpart
4. Learn and integrate knowledge from previous interactions with humans

Over the past three years the Steering Lab has worked intensively on all of the above aspects.

Project Objective: The major goal of this project is to develop an intrinsic motivation complex for an artificial conversational assistant for continuous learning and adaptation to user’s preferences and behavior covering specific conversation domain. The personal conversational assistant shall be able to express its own needs and demands and act/respond accordingly. This AI feature is key to driving human acceptance and collaboration.

Algorithms: Machine Learning (advanced natural language understanding (NLU) methods, generative models, classification, deep learning)
Data: The conversation domain will be given. All freely available data resources can be used. Data for specific domains can be provided.
Tools: Python.