EMERALD: An Integrated System of Machine Learning and Discovery Programs to Support AI Education and Experimental Research

With the rapid expansion of machine learning methods and applications, there is a strong need for computer-based interactive tools that support education in this area. The EMERALD system was developed to provide hands-on experience and an interactive demonstration of several machine learning and discovery capabilities for students in AI and cognitive science, and for AI professionals. The current version of EMERALD integrates five programs that exhibit different types of machine learning and discovery: learning rules from examples, determining structural descriptions of object classes, inventing conceptual clusterings of entities, predicting sequences of objects, and discovering equations characterizing collections of quantitative and qualitative data. EMERALD extensively uses color graphic capabilities, voice synthesis, and a natural language representation of the knowledge acquired by the learning programs. Each program is presented as a "learning robot," which has its own "personality," expressed by its icon, its voice, the comments it generates during the learning process, and the results of learning presented as natural language text and/or voice output. Users learn about the capabilities of each "robot" both by being challenged to perform some learning tasks themselves, and by creating their own similar tasks to challenge the "robot." EMERALD is an extension of ILLIAN, an initial, much smaller version that toured eight major US Museums of Science, and was seen by over half a million visitors. EMERALD's architecture allows it to incorporate new programs and new capabilities. The system runs on SUN workstations, and is available to universities and educational institutions.