Thomas R. Hinrichs

Companion cognitive systems: a step toward human-level AI

We are developing Companion Cognitive Systems, a new kind of software that can be effectively treated as a collaborator. Aside from their potential utility, we believe this effort is important because it focuses on three key problems that must be solved to achieve human-level AI: Robust reasoning and learning, interactivity, and longevity. We describe the ideas we are using to develop the first architecture for Companions: analogical processing, grounded in cognitive science for reasoning and learning, sketching and concept maps to improve interactivity, and a distributed agent architecture hosted on a cluster to achieve performance and longevity. We outline some results on learning by accumulating examples derived from our first experimental version.

Companion Cognitive Systems: Design Goals and Lessons Learned So Far

The companion cognitive architecture supports experiments in achieving human-level intelligence. This article describes seven key design goals of companions, relating them to properties of human reasoning and learning, and to engineering concerns raised by attempting to build large-scale cognitive systems. We summarize our experiences with companions in two domains: test taking and game playing.

Transfer Learning through Analogy in Games

We report on a series of transfer learning experiments in game domains, in which we use structural analogy from one learned game to speed learning of another related game. We find that a major benefit of analogy is that it reduces the extent to which the source domain must be generalized before transfer. We describe two techniques in particular, minimal ascension and metamapping, that enable analogies to be drawn even when comparing descriptions using different relational vocabularies. Evidence for the effectiveness of these techniques is provided by a large-scale external evaluation, involving a substantial number of novel distant analogs.

Interface design based on standardized task models

Producing high-quality, comprehensible user intetices is a difficult, labor-intensive process that requires experience and judgment. In this paper, we describe an approach to assisting this process by using explicit models of the user’s task to drive the interface design process. The task model helps to ensure that the resulting interf%ce directly and transparently supports the user in performing his task. By crafting a library of standardized, reusable tasks and interface constructs, we believe it is possible to capture some of the design expertise and to amortize much of the labor required for building effective user interfaces.

An interface design tool based on explicit task models

Producing high-quality, comprehensible human interfaces is a difficult, labor-intensive process that requires experience and judgment. In this paper, we describe an approach to assisting this process by using explicit models of the user’s task to drive the interface design and to serve as a functional component of the interface itself. The task model helps to ensure that the resulting interface directly and transpru-ently supports the user in performing his task, and serves as a scaffolding for providing in-context help and advice. By crafting a library of standardized, reusable tasks and interface constructs, we believe it is possible to capture some of the design expertise and to amortize much of the labor required for building effective user interfaces.

Some Limitations of Feature-Based Recognition in Case-Based Design

A crucial part of Case-Based Reasoning is retrieving cases that are similar or otherwise relevant to the problem at hand. Traditionally, this has been formulated as a problem of indexing and accessing cases based on sets of predictive features. More generally, however, we can think of retrieval as a problem of recognition. In this light, several limitations of the feature-based approach become apparent. What constitutes a feature? What makes a feature predictive? And how is retrieval possible when the structure of an input is predictive, but its components are not?

Sketching as a Modality in Intelligent Tutoring Systems

The Spatial Intelligence and Learning Center (SILC) pioneered the idea of spatial learning: improving learning about spatial concepts and using spatial concepts to facilitate learning about other domains. Spatial concepts are of significant importance in Science, Technology, Engineering and Mathematics (STEM) education including physics, chemistry, geoscience, and biology, as well as most branches of engineering. Spatial concepts are also important in the military. For example, sand tables are used to understand the terrain, since calculating Line of Sight (LOS) for cover and concealment on the battlefield can mean the difference between life or death. In addition, land navigation using topographic maps is a critical skill that all members of the military are required to obtain.

Improved Performance Support through an Integrated Task-Based Video Case Library

Case-based retrieval and other decision support systems typically exist separately from the tools and tasks they support. Users are required to initiate searches and identify target case features manually, and as a result the systems are not used to their full extent. We describe an approach to integrating an ASK system--a type of video case library--with a performance support tool. This approach uses model-based task tracking to retrieve cases relevant to how a user is performing a task, not just to the artifacts that are created during the process.