Neural networks and pattern recognition in human-computer interaction

Abstract

This paper reports on the activities of the workshop held on Sunday 28th April at the CHI'91 conference. Participants were there to discuss different ideas, methods and approaches to using pattern recognition in human-computer interaction.The workshop aimed to bring together researchers using novel methodologies, such as neural networks, in HCI applications, as well as practitioners using alternative or more traditional methods to perform pattern recognition tasks in HCI. The intention was to explore the scope and limitations of each type of approach and its requirements, for example in terms of representation and resources. The workshop considered the relationships between the different approaches and the possibility of developing hybrid methodologies to resolve HCI problems.Researchers working with both traditional and novel pattern recognition methods that have applications to human-computer interaction, and those with strong views either way, submitted position statements outlining their interest and viewpoints. Their research results are summarised in this report; in addition, the discussions on methods, on how the work reported interrelates, and on future areas of interest are presented. Major results from the use of neural network systems and other pattern recognition systems in the interface are presented, with application areas ranging from the interpretation of gestural input to the automatic determination of user task. Fuller details of the research can be found in a book based on the proceedings[2].