No abstract

We describe a cognitive architecture for creating more robust intelligent systems by executing hybrids of algorithms based on different computational formalisms. The architecture is motivated by the belief that (1) most existing computational methods often exhibit some of the characteristics desired of intelligent systems at the cost of other desired characteristics and (2) a system exhibiting robust intelligence can be designed by implementing hybrids of these computational methods. The main obstacle to this approach is that the various relevant computational methods are based on data structures and algorithms that are very difficult to integrate into one system. We describe a new method of executing hybrids of algorithms using the focus of attention of multiple modules. This approach has been embodied in the Polyscheme cognitive architecture. Systems based on Polyscheme can integrate reactive robotic controllers, logical and probabilistic inference algorithms, frame-based formalisms and sensor-processing algorithms into one system. Existing applications involve human-robot interaction, heterogeneous information retrieval and natural language understanding. Systems built using Polyscheme demonstrate that algorithmic hybrids implemented using a focus of attention can (1) exhibit more characteristics of intelligence than individual computational methods alone and (2) deal with problems that have formerly been beyond the reach of synthetic computational intelligence.

The theory that human cognition proceeds through mental simulations, if true, would provide a parsimonious explanation of how the mechanisms of reasoning and problem solving integrate with and develop from mechanisms underlying forms of cognition that occur earlier in evolution and development. However, questions remain about whether simulation mechanisms are powerful enough to exhibit human-level reasoning and inference. In order to investigate this issue, we show that it is possible to characterize some of the most powerful modern artificial intelligence algorithms for logical and probabilistic inference as methods of simulating alternate states of the world. We show that a set of specific human perceptual mechanisms, even if not implemented using mechanisms described in artificial intelligence, can nevertheless perform the same operations as those algorithms. Although this result does not demonstrate that simulation theory is true, it does show that whatever mechanisms underlie perception have at least as much power to explain non-perceptual human reasoning and problem solving as some of the most powerful known algorithms.

No abstract

Satisfiability (SAT) testing methods have been used effectively in many inference, planning and constraint satisfaction tasks and thus have been considered a contribution towards artificial general intelligence. However, since SAT constraints are defined over atomic propositions, domains with state variables that change over time can lead to extremely large search spaces. This poses both memory-and time-efficiency problems for existing SAT algorithms. In this paper, we propose to address these problems by introducing a language that encodes the temporal intervals over which relations occur and an integrated system that satisfies constraints formulated in this language. Temporal intervals are presented as a compressed method of encoding time that results in significantly smaller search spaces. However, intervals cannot be used efficiently without significant modifications to traditional SAT algorithms. Using the Polyscheme cognitive architecture, we created a system that integrates a DPLL-like SATsolving algorithm with a rule matcher in order to support intervals by allowing new constraints and objects to be lazily instantiated throughout inference. Our system also includes constraint graphs to compactly store information about temporal and identity relationships between objects. In addition, a memory retrieval subsystem was utilized to guide inference towards minimal models in common sense reasoning problems involving time and change. We performed two sets of evaluations to isolate the contributions of the system"s individual components. These tests demonstrate that both the ability to add new objects during inference and the use of smart memory retrieval result in a significant increase in performance over pure satisfiability algorithms alone and offer solutions to some problems on a larger scale than what was possible before.