Our goal is for an autonomous learning agent to acquire the knowledge that serves as the foundations of common sense from its own experience without outside guidance. This requires the agent to (1) learn the structure of its own sensors and effectors; (2) learn a model of space around itself; (3) learn to move effectively in that space; (4) identify and describe objects, as distinct from the static environment; (5) learn and represent actions for affecting those objects, including preconditions and postconditions, and so on. We will provide examples of progress we have made, and the roadmap we envision for future research.

Agent adaptability — the ability of agents to change behavioral strategies when it is beneficial to do so — is presumed to be an important part of the robustness of complex adaptive systems (CAS). But, determining when changing behaviors is advantageous for agents has proven quite challenging in CAS research, as sometimes behavioral change is necessary, but other times it can impose costs that exceed benefits. I present the results from experiments using an agent-based model (ABM) designed to discover thresholds after which behavioral flexibility leads to improved societal-level outcomes in groups of agents in dynamic environments. The first major result is that there are thresholds in both levels of flexibility in agent behavior and in levels of turbulence in the environment below and above which there are marked differences in utility gains for agents. In particular, relatively high flexibility leads to lower overall utility scores, as well as, surprisingly, decreased diversity and increased inequality between agents. The second result is that at very high levels of environmental turbulence, the effects of the environment alone on agent utility overshadow any benefits to agents from flexible behavior strategies. This suggests, counter-intuitively, that the best strategy for agents in very dynamic environments is simply to keep behavior constant. The third major result is that there is an interaction between agent behavior and the environment: high flexibility of other agents can effectively make an environment more "dynamic", which just fuels more flexibility, and leads to a scramble between different strategies with no utility gain. A final theoretical contribution of the paper is that the model is able to show drawbacks to flexibility without relying on costs to changing behavior, as is done in much of the literature on strategy change.

Modern mobile robots navigate uncertain environments using complex compositions of camera, laser, and sonar sensor data. Manual calibration of these sensors is a tedious process that involves determining sensor behavior, geometry and location through model specification and system identification. Instead, we seek to automate the construction of sensor model geometry by mining uninterpreted sensor streams for regularities. Manifold learning methods are powerful techniques for deriving sensor structure from streams of sensor data. In recent years, the proliferation of manifold learning algorithms has led to a variety of choices for autonomously generating models of sensor geometry. We present a series of comparisons between different manifold learning methods for discovering sensor geometry for the specific case of a mobile robot with a variety of sensors. We also explore the effect of control laws and sensor boundary size on the efficacy of manifold learning approaches. We find that "motor babbling" control laws generate better geometric sensor maps than mid-line or wall following control laws and identify a novel method for distinguishing boundary sensor elements. We also present a new learning method, sensorimotor embedding, that takes advantage of the controllable nature of robots to build sensor maps.

To help realize the potential of complex systems models we need new measures appropriate for capturing processes that exhibit feedback, nonlinearity, heterogeneity, and emergence. As part of a larger research project encompassing several categories of dynamical properties this paper provides formal and general definitions of tipping point-related phenomena. For each tipping concept this paper provides a probabilistic definition derived from a Markov model representation. We start with the basic features of Markov models and definitions of the foundational concepts of system dynamics. Then several tipping point-related concepts are described, defined, measured, and illustrated with a simplified graphical example. The paper finishes with several branches of future work involving new measures for complex systems and the fusion of research domains.

Our past work has investigated the use of the Cognitive Model Software Development Kit (SDK) for creating the cognitive models that underlie model-tracing Cognitive Tutors. Though successful at increasing the number of people who could author such a cognitive model, for certain kinds of situations the Cognitive Model SDK proved cumbersome. The present work discusses a new authoring system, xPST, that allows an example-based tutor to be built on top of existing software.

Over the last few years, a number of areas of natural language processing have begun applying graph-based techniques. These include, among others, text summarization, syntactic parsing, word-sense disambiguation, ontology construction, sentiment and subjectivity analysis, and text clustering. In this paper, we present some of the most successful graph-based representations and algorithms used in language processing and try to explain how and why they work.

This introduction to AI Magazine's Special Issueon Networks and AI summarizes the seven articles in thespecial issue by characterizing the nature of thenetworks that are the focus of each of the papers.A short tutorial on graph theory and network structuresis included for those less familiar with the topic.

Over the last few years, a number of areas of natural language processing have begun applying graph-based techniques. These include, among others, text summarization, syntactic parsing, word-sense disambiguation, ontology construction, sentiment and subjectivity analysis, and text clustering. In this paper, we present some of the most successful graph-based representations and algorithms used in language processing and try to explain how and why they work.