论文信息 - Representing conceptual structures in a neural network

Representing conceptual structures in a neural network

Abstract : DUCS is an experimental multi-level architecture that provides a distributed representation for frame-like concept structures. The ultimate goal of this research is to develop a powerful short term memory that can construct and manipulate concepts rapidly. Such a memory could plan an important role in neural network models of higher level cognitive phenomena such as language understanding or problem solving. Keywords: Artificial intelligence, Machine learning, Connectionism, Short-term memory, Distributed representation, Frames.

David S. Touretzky | D. Touretzky

[1] David S. Touretzky,et al. Symbol Structures in Connectionist Networks: Five Properties and Two Architectures , 1987, COMPCON.

[2] Tj Sejnowski,et al. Skeleton filters in the brain , 2014 .

[3] James L. McClelland,et al. Mechanisms of Sentence Processing: Assigning Roles to Constituents of Sentences , 1986 .

[4] J J Hopfield,et al. Neural networks and physical systems with emergent collective computational abilities. , 1982, Proceedings of the National Academy of Sciences of the United States of America.

[5] Geoffrey E. Hinton. A Parallel Computation that Assigns Canonical Object-Based Frames of Reference , 1981, IJCAI.

[6] David S. Touretzky,et al. A distributed connectionist representation for concept structures , 1987 .

[7] Donald A. Norman,et al. The perception of multiple objects: a parallel, distributed processing approach , 1987 .

[8] Geoffrey E. Hinton,et al. A Distributed Connectionist Production System , 1988, Cogn. Sci..