An Architecture for Adaptive Algorithmic Hybrids

Cassimatis, Nicholas L.; Bignoli, Perrin; Bugajska, Magdalena D.; Dugas, Scott; Kurup, Unmesh; Murugesan, Arthi; Bello, Paul

doi:10.1109/tsmcb.2009.2033262

Cited by 16 publications

(4 citation statements)

References 23 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The literature on human attention is of course full of insights into how people order their attention and thus a place to look for potential variable ordering heuristics. In preliminary work (Cassimatis et al 2007), we have used the Polyscheme cognitive architecture to implement these algorithms as simulations. The work suggests that such an integration between reasoning algorithms and ''lower-level'' mechanisms can lead these algorithms to exhibit abilities that had not previously been possible.…”

Section: Discussionmentioning

confidence: 99%

Reasoning as simulation

2009

Self Cite

View full text Add to dashboard Cite

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.

show abstract

Section: Discussionmentioning

confidence: 99%

Reasoning as simulation

2009

Self Cite

View full text Add to dashboard Cite

show abstract

“…Positive pre-conditions without which the effect of the model is reliably entailed are deemed irrelevant: prediction monitors will repeatedly decrease their reliability until deletion. When presented with a goal, M 0 outputs a sub-goal (iM 0 (y z)) targeting its own (forward) operation -this 4 Assumptions are not essential for the present discussion and will not be detailed further.…”

Section: Modelsmentioning

confidence: 99%

Anytime Bounded Rationality

Nivel

Þórisson

Steunebrink

et al. 2015

Artificial General Intelligence

View full text Add to dashboard Cite

Abstract. Dependable cyber-physical systems strive to deliver anticipative, multi-objective performance anytime, facing deluges of inputs with varying and limited resources. This is even more challenging for life-long learning rational agents as they also have to contend with the varying and growing know-how accumulated from experience. These issues are of crucial practical value, yet have been only marginally and unsatisfactorily addressed in AGI research. We present a value-driven computational model of anytime bounded rationality robust to variations of both resources and knowledge. It leverages continually learned knowledge to anticipate, revise and maintain concurrent courses of action spanning over arbitrary time scales for execution anytime necessary. IntroductionKey among the properties mission-critical systems call for is anytime control -the capability of a controller to produce control inputs whenever necessary, despite the lack of resources, trading quality for responsiveness [3,5]. Any practical AGI is constrained by a mission, its own architecture, and limited resources including insufficient time/memory to process all available inputs in order to achieve the full extent of its goals when it matters. Moreover, unlike fully hand-crafted cyber-physical systems, AGIs should handle underspecified dynamic environments, with no other choice but to learn their know-how, possibly throughout their entire lifetime. The challenge of anytime control thus becomes broader as, in addition to resource scarcity, it must encompass inevitable variations of completeness, consistency, and accuracy of the learned programs from which decisions are derived. We address the requirement of delivering anticipative, multi-objective and anytime performance from a varying body of knowledge. A system must anticipate its environment for taking appropriate action -a controller that does not can only react after the facts and "lag behind the plant". Predictions and sub-goals must be produced concurrently: (a) since achieving goals needs predictions, the latter must be up to date; (b) a complex environment's state transitions can never be predicted entirely: the most interesting ones are those that pertain to the achievements of the system's goals, so these must be up to date when predictions are generated. A system also needs to achieve multiple concurrent goals to reach states that can only be obtained using several independent yet temporally correlated and/or co-dependent courses of action

show abstract

“…The elements of this list are not original (see (Cassimatis, 2010) and (Shanahan, 1997)) or exhaustive. They are merely illustrative examples:…”

Section: The State Of the Sciencementioning

confidence: 99%

Artificial Intelligence and Cognitive Modeling Have the Same Problem

Cassimatis¹

2012

Atlantis Thinking Machines

View full text Add to dashboard Cite

Cognitive modelers attempting to explain human intelligence share a puzzle with artificial intelligence researchers aiming to create computers that exhibit human-level intelligence: how can a system composed of relatively unintelligent parts (such as neurons or transistors) behave intelligently? I argue that although cognitive science has made significant progress towards many of its goals, that solving the puzzle of intelligence requires special standards and methods in addition to those already employed in cognitive science. To promote such research, I suggest creating a subfield within cognitive science called intelligence science and propose some guidelines for research addressing the intelligence puzzle. The Intelligence ProblemCognitive scientists attempting to fully understand human cognition share a puzzle with artificial intelligence researchers aiming to create computers that exhibit human-level intelligence: how can a system composed of relatively unintelligent parts (say, neurons or transistors) behave intelligently? 2.1.1 Naming the problem I will call the problem of understanding how unintelligent components can combine to generate human-level intelligence the intelligence problem; the endeavor to understand how the human brain embodies a solution to this problem understanding human intelligence; and the project of making computers with human-level intelligence human-level artificial intelligence. 11

show abstract

An Architecture for Adaptive Algorithmic Hybrids

Cited by 16 publications

References 23 publications

Reasoning as simulation

Reasoning as simulation

Anytime Bounded Rationality

Artificial Intelligence and Cognitive Modeling Have the Same Problem

Contact Info

Product

Resources

About