This paper presents an approach to learning from situated, interactive tutorial instruction within an ongoing agent. Tutorial instruction is a flexible (and thus powerful) paradigm for teaching tasks because it allows an instructor to communicate whatever types of knowledge an agent might need in whatever situations might arise. To support this flexibility, however, the agent must be able to learn multiple kinds of knowledge from a broad range of instructional interactions. Our approach, called situated explanation, achieves such learning through a combination of analytic and inductive techniques. It combines a form of explanation-based learning that is situated for each instruction with a full suite of contextually guided responses to incomplete explanations. The approach is implemented in an agent called Instructo-Soar that learns hierarchies of new tasks and other domain knowledge from interactive natural language instructions. Instructo-Soar meets three key requirements of flexible...

The origins of the Soar architecture can be traced back to the seminal research of Allen Newell and Herbert Simon on symbol systems, heuristic search, goals, problem spaces, and production systems. Since its official inception in 1982, Soar has evolved through six major releases, as both an AI architecture and as the basis for a unified theory of cognition. This paper traces this evolutionary path, starting with Soar's intellectual roots, and then proceeding through the stages defined by the six major system releases. Each stage is characterized with respect to a hierarchy of four levels of analysis: the knowledge level, the problem space level, the symbolic architecture level, and the implementation level.

Many researchers have noted the importance of combining inductive and analytical learning, yet we still lack combined learning methods that are effective in practice. We present here a learning method that combines explanation-based learning from a previously learned approximate domain theory, together with inductive learning from observations. This method, called explanation-based neural network learning (EBNN), is based on a neural network representation of domain knowledge. Explanations are constructed by chaining together inferences from multiple neural networks. In contrast with symbolic approaches to explanation-based learning which extract weakest preconditions from the explanation, EBNN extracts the derivatives of the target concept with respect to the training example features. These derivatives summarize the dependencies within the explanation, and are used to bias the inductive learning of the target concept. Experimental results on a simulated robot control task show that E...

In contrast to current intelligent systems, which must be laboriously programmed for each task they are meant to perform, instructable agents can be taught new tasks and associated knowledge. This thesis presents a general theory of learning from tutorial instruction and its use to produce an instructable agent. Tutorial instruction is a particularly powerful form of instruction, because it allows the instructor to communicate whatever kind of knowledge a student needs at whatever point it is needed. To exploit this broad flexibility, however, a tutorable agent must support a full range of interaction with its instructor to learn a full range of knowledge. Thus, unlike most machine learning tasks, which target deep learning of a single kind of knowledge from a single kind of input, tutorability requires a breadth of learning from a broad range of instructional interactions. The theory of learning from tutorial...

This paper describes an approach to learning a simple indoor robot navigation task through trial-and-error. A mobile robot, equipped with visual, ultrasonic and laser sensors, learns to servo to a designated target object. In less than ten minutes of operation time, the robot is able to navigate to a marked target object in an office environment. The central learning mechanism is the explanationbased neural network learning algorithm (EBNN). EBNN initially learns function purely inductively using neural network representations. With increasing experience, EBNN employs domain knowledge to explain and to analyze training data in order to generalize in a more knowledgeable way. Here EBNN is applied in the context of reinforcement learning, which allows the robot to learn control using dynamic programming. Keywords: explanation-based learning, mobile robots, machine learning, navigation, neural networks, perception 1 Introduction Throughout the last decades, the field of robotics has prod...

This paper describes CES, a proto-type of a new programming language for robots and other embedded systems, equipped with sensors and actuators. CES contains two new ideas, currently not found in other programming languages: support of computing with uncertain information, and support of adaptation and teaching as a means of programming. These innovations facilitate the rapid development of software for embedded systems, as demonstrated by a mobile robot application.

In previous work we described a knowledge-intensive inductive learning algorithm called abductive explanation-based learning (A-EBL) that uses background knowledge to improve the performance of a concept learner. A disadvantage of A-EBL is that it is not incremental. This paper describes an alternative learning algorithm called IA-EBL that learns incrementally; IA-EBL replaces the set-cover based learning algorithm of A-EBL with an extension of a perceptron learning algorithm. However, IA-EBL is in most other respects comparable to A-EBL, except that the output of the learning system can no longer be easily expressed as a logical theory. In this paper, IA-EBL is described, analyzed according to Littlestone's model of mistake-bounded learnability, and finally compared experimentally to A-EBL. IA-EBL is shown to provide order-ofmagnitude speedups over A-EBL in two domains when used in an incremental setting.

this paper include: Both inductive and analytical learning mechanisms will be needed to cover the range of learning exhibited by humans and other intelligent systems. Analytical mechanisms are required in order to scale up to learning complex concepts, and to handle situations in which available training data is limited. Inductive mechanisms are required in order to learn in situations where prior knowledge is incomplete or incorrect

What is the match cost of adding a new rule to a production system (rule-based system)? Two conflicting views have emerged. Research in EBL indicates that learned rules add to the match cost of a production system. Thus, as the production system size increases with learning, the match cost will also increase. There is much data in the literature to support this phenomenon. On the contrary, researchers in parallel production systems have concluded that the match effort in a production system is limited, independent of the size of the production system. Thus, an increase in the size of the production system will not lead to an increase in the match cost. There is much data to support this phenomenon as well. In this paper, we point out these contradictory views of production match in the two research communities. A direct analysis of these conflicting views is difficult, since the two communities have worked with vastly different systems. Therefore, we have developed some large production systems in Soar, to analyze the situation within a common framework. This common framework narrows down the possible causes for this conflict, and raises important questions for future work.

A consistent pattern of changes in the 31 P MR spectra of normal premenopausal breast during the menstrual cycle has been observed. These encouraging preliminary data suggest that magnetic resonance spectroscopy (MRS) may have a role in monitoring hormone-dependent...