Anchoring Knowledge in Interaction: Towards a Harmonic Subsymbolic/Symbolic Framework and Architecture of Computational Cognition
We outline a proposal for a research program leading to a new paradigm, architectural framework, and prototypical implementation, for the cognitively inspired anchoring of an agent’s learning, knowledge formation, and higher reasoning abilities in real-world interactions: Learning through interaction in real-time in a real environment triggers the incremental accumulation and repair of knowledge that leads to the formation of theories at a higher level of abstraction. The transformations at this higher level filter down and inform the learning process as part of a permanent cycle of learning through experience, higher-order deliberation, theory formation and revision.
The envisioned framework will provide a precise computational theory, algorithmic descriptions, and an implementation in cyber-physical systems, addressing the lifting of action patterns from the subsymbolic to the symbolic knowledge level, effective methods for theory formation, adaptation, and evolution, the anchoring of knowledge-level objects, real-world interactions and manipulations, and the realization and evaluation of such a system in different scenarios. The expected results can provide new foundations for future agent architectures, multi-agent systems, robotics, and cognitive systems, and can facilitate a deeper understanding of the development and interaction in human-technological settings.
KeywordsTransfer Learning Restrict Boltzmann Machine Analogical Transfer Deep Network Chord Progression
Unable to display preview. Download preview PDF.
- 1.Arnold, L., Paugam-Moisy, H., Sebag, M.: Unsupervised layer-wise model selection in deep neural networks. In: Proceedings of ECAI 2010: 19th European Conference on Artificial Intelligence, pp. 915–920. IOS Press (2010)Google Scholar
- 3.Bundy, A.: The interaction of representation and reasoning. Proceedings of the Royal Society A: Mathematical, Physical and Engineering Sciences 469(2157) (2013)Google Scholar
- 5.Coradeschi, S., Saffiotti, A.: Anchoring symbols to sensor data: preliminary report. In: Proceedings of the 17th AAAI Conference, pp. 129–135. AAAI Press (2000)Google Scholar
- 7.De Penning, H.L.H., Garcez, A.S.D., Lamb, L.C., Meyer, J. J. C.: A Neural-symbolic cognitive agent for online learning and reasoning. In: Proceedings of the 22nd International Joint Conference on Artificial Intelligence, pp. 1653–1658. AAAI Press (2011)Google Scholar
- 8.Diaconescu, R.: Institution-independent Model Theory. Birkhäuser, 1st edn. (2008)Google Scholar
- 11.Gärdenfors, P.: Conceptual Spaces: The Geometry of Thought. MIT Press (2000)Google Scholar
- 12.Gkaniatsou, A., Bundy, A., Mcneill, F.: Towards the automatic detection and correction of errors in automatically constructed ontologies. In: 8th International Conference on Signal Image Technology and Internet Based Systems 2012, pp. 860–867 (2012)Google Scholar
- 15.Jaeger, H.: Controlling recurrent neural networks by conceptors. arXiv (2014), 1403.3369v1 [cs.CV] (March 13, 2014)Google Scholar
- 16.LeBlanc, K., Saffiotti, A.: Cooperative anchoring in heterogeneous multi-robot systems. In: 2008 IEEE International Conference on Robotics and Automation, pp. 3308–3314 (2008)Google Scholar
- 18.Lovett, A., Forbus, K., Usher, J.: A structure-mapping model of raven’s progressive matrices. In: 32nd Annual Meeting of the Cognitive Science Society, pp. 2761–2766 (2010)Google Scholar
- 19.Marr, D.: Vision. A Computational Investigation into the Human Representation and Processing of Visual Information. W. H. Freeman and Company (1982)Google Scholar
- 20.McLure, M., Friedman, S., Forbus, K.: Learning concepts from sketches via analogical generalization and near-misses. In: 32nd Annual Meeting of the Cognitive Science Society, pp. 1726–1731 (2010)Google Scholar