Lee Spector

Ontology-based Web agents

This paper describes SHOE, a set of Simple HTML Ontology Extensions which allow World-Wide Web authors to annotate their pages with semantic knowledge such as “I am a graduate student” or “This person is my graduate advisor”. These annotations are expressed in terms of ontological knowledge which can be generated by using or extending standard ontologies available on the Web. This makes it possible to ask Web agent queries such as “Find me all graduate students in Maryland who are working on a project funded by DoD initiative 123-4567”, instead of simplistic keyword searches enabled by current search engines. We have also developed a web-crawling agent, Expos´ e, which interns SHOE knowledge from web documents, making these kinds queries a reality.

Genetic Programming and Autoconstructive Evolution with the Push Programming Language

Push is a programming language designed for the expression of evolving programs within an evolutionary computation system. This article describes Push and illustrates some of the opportunities that it presents for evolutionary computation. Two evolutionary computation systems, PushGP and Pushpop, are described in detail. PushGP is a genetic programming system that evolves Push programs to solve computational problems. Pushpop, an “autoconstructive evolution” system, also evolves Push programs but does so while simultaneously evolving its own evolutionary mechanisms.

Emergence of Collective Behavior in Evolving Populations of Flying Agents

We demonstrate the emergence of collective behavior in two evolutionary computation systems, one an evolutionary extension of a classic (highly constrained) flocking algorithm and the other a relatively un-constrained system in which the behavior of agents is governed by evolved computer programs. The first system demonstrates the evolution of a form of multicellular organization, while the second demonstrates the evolution of a form of altruistic food sharing. In this article we describe both systems in detail, document the emergence of collective behavior, and argue that these systems present new opportunities for the study of group dynamics in an evolutionary context. We also provide a brief overview of the breve simulation environment in which the systems were produced, and of breve’s facilities for the rapid, exploratory development of visualization strategies for artificial life.

The Push3 execution stack and the evolution of control

The Push programming language was developed for use in genetic and evolutionary computation systems, as the representation within which evolving programs are expressed. It has been used in the production of several significant results, including results that were awarded a gold medal in the Human Competitive Results competition at GECCO-2004. One of Pushs attractive features in this context is its transparent support for the expression and evolution of modular architectures and complex control structures, achieved through explicit code self-manipulation. The latest version of Push, Push3, enhances this feature by permitting explicit manipulation of an execution stack that contains the expressions that are queued for execution in the interpreter. This paper provides a brief introduction to Push and to execution stack manipulation in Push3. It then presents a series of examples in which Push3 was used with a simple genetic programming system (PushGP) to evolve programs with non-trivial control structures.

Finding a better-than-classical quantum AND/OR algorithm using genetic programming

This paper documents the discovery of a new, better-than-classical quantum algorithm for the depth-two AND/OR tree problem. We describe the genetic programming system that was constructed specifically for this work, the quantum computer simulator that is used to evaluate the fitness of evolving quantum algorithms, and the newly discovered algorithm.

Wolf-pack (Canis lupus) hunting strategies emerge from simple rules in computational simulations

We have produced computational simulations of multi-agent systems in which wolf agents chase prey agents. We show that two simple decentralized rules controlling the movement of each wolf are enough to reproduce the main features of the wolf-pack hunting behavior: tracking the prey, carrying out the pursuit, and encircling the prey until it stops moving. The rules are (1) move towards the prey until a minimum safe distance to the prey is reached, and (2) when close enough to the prey, move away from the other wolves that are close to the safe distance to the prey. The hunting agents are autonomous, interchangeable and indistinguishable; the only information each agent needs is the position of the other agents. Our results suggest that wolf-pack hunting is an emergent collective behavior which does not necessarily rely on the presence of effective communication between the individuals participating in the hunt, and that no hierarchy is needed in the group to achieve the task properly.

Damage to the prefrontal cortex leads to decomposition of structured event complexes

Head injury often results in focal prefrontal lobe lesions that may account for specific posttraumatic cognitive and personality changes. These changes are briefly reviewed as well as some of the theoretical constructs used to describe the kinds of cognitive processes that are stored in prefrontal cortex. A new framework for describing prefrontal cortex function is then introduced that provides a rational and theoretically rich cognitive architecture that lends itself to hypothesis testing.

PARKA: parallel knowledge representation on the Connection Machine

Abstract Semantic network systems are a common knowledge representation paradigm used in contemporary AI research. Researchers have become increasingly discouraged, however, with the performance of these systems. As networks grow performance suffers, often to an unacceptable extent. The representations for many AI domains are so large that existing systems cannot execute general inferences on them in an acceptable amount of time-serial representation systems are computationally ineffective for large domains. This is one of the reasons that, in the last few years, researchers have been turning their attention, away from traditional symbolic approaches, toward neural networks and parallel connectionist paradigms. We have been developing PARKA, a symbolic, semantic network knowledge representation system that takes advantage of the CM′s massive parallelism. Our primary motivation is to demonstrate that massive parallelism can be used to provide the run-time improvements necessary to bring large AI systems into realistic applications requiring rapid response time. PARKA′s design allows it to evaluate rapidly enough for everyday use many inferences that are computationally ineffective in serial systems. In this paper we describe the design and implementation of PARKA. We explain our motivations for developing a representation system on the Connection Machine and show how PARKA exploits its design to optimize performance. We show that PARKA effects in O(d) time inferences for which serial systems require O(Bd) time, where d is the network depth and B is the average network branchout. Timings of the system on random networks are also presented. We show that PARKA is capable of effecting top-down inheritance inferences on 16,000-node networks in well under a second.

Solving Uncompromising Problems With Lexicase Selection

We describe a broad class of problems, called “uncompromising problems,” which are characterized by the requirement that solutions must perform optimally on each of many test cases. Many of the problems that have long motivated genetic programming research, including the automation of many traditional programming tasks, are uncompromising. We describe and analyze the recently proposed “lexicase” parent selection algorithm and show that it can facilitate the solution of uncompromising problems by genetic programming. Unlike most traditional parent selection techniques, lexicase selection does not base selection on a fitness value that is aggregated over all test cases; rather, it considers test cases one at a time in random order. We present results comparing lexicase selection to more traditional parent selection methods, including standard tournament selection and implicit fitness sharing, on four uncompromising problems: 1) finding terms in finite algebras; 2) designing digital multipliers; 3) counting words in files; and 4) performing symbolic regression of the factorial function. We provide evidence that lexicase selection maintains higher levels of population diversity than other selection methods, which may partially explain its utility as a parent selection algorithm in the context of uncompromising problems.

Emergence of collective behavior in evolving populations of flying agents

We demonstrate the emergence of collective behavior in two evolutionary computation systems, one an evolutionary extension of a classic (highly constrained) flocking algorithm and the other a relatively un-constrained system in which the behavior of agents is governed by evolved computer programs. We describe the systems in detail, document the emergence of collective behavior, and argue that these systems present new opportunities for the study of group dynamics in an evolutionary context.