The Metacognitive Loop and Reasoning about Anomalies

Schmill, Matthew D.; Anderson, Margaret; Fults, Scott; Josyula, Darsana P.; Oates, Tim; Perlis, Donald; Shahri, Hamid Haidarian; Wilson, Shomir; Wright, Dean

doi:10.7551/mitpress/9780262014809.003.0012

Cited by 14 publications

(15 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Second, in both cases there is a clear progression as a larger percentage of the window slides into the anomalous region. At first GNG maps to nodes that are slightly anomalous (3,4), then nodes that are moderately anomalous (5,6) and finally those that are fully anomalous (7,8) as the window's back edge enters the anomaly. This pattern is reversed as the window slides out of the anomalous section, showing that the network has generated anomaly prototypes which reflect intensity as well as type.…”

Section: Resultsmentioning

confidence: 99%

“…Several iterations of MCL have been developed, including domainspecific versions that guide agents to improved performance in Q-learning [2] and natural language dialog [3] systems. A domain-independent version of MCL based on a Bayesian net was also developed [4] and used to guide a simulated mars rover and for other tasks.…”

Section: Introduction and Related Workmentioning

confidence: 99%

See 1 more Smart Citation

Symbolic Anomaly Detection and Assessment Using Growing Neural Gas

Paisner

Perlis

Cox

2013

2013 IEEE 25th International Conference on Tools With Artificial Intelligence

Self Cite

View full text Add to dashboard Cite

Metacognitive architectures provide one solution to the brittleness problem for agents operating in complex, changing environments. The Metacognitive Loop, in which a system notes an anomaly, assesses the problem and guides a solution, is one form of such an architecture. This paper extends prior work on implementing the note phase of this process in symbolic planning domains using the A-distance. This extension uses a growing neural gas algorithm to construct a network which represents various normal and anomalous states. Testing shows that this technique allows for improved detection of anomalies in the note phase as well as categorization of anomalies by severity and type in the assess phase.

show abstract

Section: Resultsmentioning

confidence: 99%

Section: Introduction and Related Workmentioning

confidence: 99%

Symbolic Anomaly Detection and Assessment Using Growing Neural Gas

Paisner

Perlis

Cox

2013

2013 IEEE 25th International Conference on Tools With Artificial Intelligence

Self Cite

View full text Add to dashboard Cite

show abstract

“…For instance, in the approach proposed by Schmill et al (2010), a metacognitive loop (MCL) layered architecture is described, which introduces also a metacognitive layer (similar to our second layer, i.e. the action selection -RL layer) aiming to provide robustness to autonomous systems in the presence of unforeseen perturbations.…”

Section: Journal Of Experimental and Theoretical Artificial Intelligencementioning

confidence: 99%

Model-free learning on robot kinematic chains using a nested multi-agent topology

Karigiannis

Tzafestas

2015

Journal of Experimental & Theoretical Artificial Intelligen

View full text Add to dashboard Cite

This paper proposes a model-free learning scheme for the developmental acquisition of robot kinematic control and dexterous manipulation skills. The approach is based on a nested-hierarchical multi-agent architecture that intuitively encapsulates the topology of robot kinematic chains, where the activity of each independent degree-of-freedom (DOF) is finally mapped onto a distinct agent. Each one of those agents progressively evolves a local kinematic control strategy in a game-theoretic sense, that is, based on a partial (local) view of the whole system topology, which is incrementally updated through a recursive communication process according to the nested-hierarchical topology. Learning is thus approached not through demonstration and training but through an autonomous self-exploration process. A fuzzy reinforcement learning scheme is employed within each agent to enable efficient exploration in a continuous state -action domain. This paper constitutes in fact a proof of concept, demonstrating that global dexterous manipulation skills can indeed evolve through such a distributed iterative learning of local agent sensorimotor mappings. The main motivation behind the development of such an incremental multi-agent topology is to enhance system modularity, to facilitate extensibility to more complex problem domains and to improve robustness with respect to structural variations including unpredictable internal failures. These attributes of the proposed system are assessed in this paper through numerical experiments in different robot manipulation task scenarios, involving both single and multi-robot kinematic chains. The generalisation capacity of the learning scheme is experimentally assessed and robustness properties of the multi-agent system are also evaluated with respect to unpredictable variations in the kinematic topology. Furthermore, these numerical experiments demonstrate the scalability properties of the proposed nested-hierarchical architecture, where new agents can be recursively added in the hierarchy to encapsulate individual active DOFs. The results presented in this paper demonstrate the feasibility of such a distributed multi-agent control framework, showing that the solutions which emerge are plausible and near-optimal. Numerical efficiency and computational cost issues are also discussed.

show abstract

“…The design of systems based on available metacognitive models is difficult, since some models are theoretical (Cox, Oates, & Perlis, 2011;Nelson & Narens, 1990) and the computational models (Alonso, Arnold, & Havasi, 2010;Kennedy, 2010;Schmill et al, 2011) do not provide formalisms of software engineering methodologies that allow systematic development. The variety of existing approaches to computational design of metacognitive processes delays the selection process of the design model for any application.…”

Section: Introductionmentioning

confidence: 99%