G.A. Carpenter and S. Grossberg. Neural dynamics of category learning and recognition: Attention, memory consolidation, and amnesia. In J. Davis, R. Newburgh, and E. Wegman, editors, Brain Structure, Learning, and Memory, AAAS Symposium Series, pages 239--286, 1986.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....group) and intensional (what features best describe each group) descriptions of unlabeled data. To explore this idea we present an artificial neural network (ANN) architecture, using as building blocks two well know models: the ART1 network, from the Adaptive Resonance Theory family of ANNs [4], and the Combinatorial Neural Model (CNM) proposed by Machado ( 11] and [12] Both models satisfy one important desiderata for data mining, learning in just one pass of the database. Moreover, CNM, the intensional part of the architecture, allows one to obtain rules directly from its ....

....data. By better descriptions we mean obtaining intensional (what are the main characteristics of each class) descriptions, beyond the extensional (what objects are members of each class) ones, usually provided. We explore our idea with a hybrid architecture, based into two well know models: ART1 [4], used for cluster binary data, and CNM ( 11] and [12] used to map the input space in the formed classes. Both ANNs present an important characteristic to support data mining: they learn in just one pass of the entire data set. The model is illustrated by Figure 1. Fig. 1. Describing ....

[Article contains additional citation context not shown here]

Carpenter, G. and Grossberg, S. Neural Dynamics of Category Learning and Recognition: Attention, Memory, Consolidation, and Amnesia. In: Joel L. Davis (ed.), Brain structure, learning, and memory. AAAS Symposia Series, Boulder, CO: Westview Press, 1988. p.233-287.

....does not match any stored pattern, a new category is created by storing a pattern that is like the input vector. Once a stored pattern is found that matches the input vector within a specified tolerance (the vigilance) that pattern is adjusted (trained) to make it even more like the input vector [CG86, CG87, Gro87, Was89, Zur92] The operation of the ART1 algorithm can be presented by six steps as follows. The presentation has been adopted from Zurada [Zur92] In the algorithm X is the input vector, which consists of binary values x i (i = 1, n) W is the bottom top processing weight ....

Carpenter G and Grossberg S: Neural dynamics of category learning and recognition: Attention, memory consolidation, and amnesia. In: Davis J, Newburgh R and Wegman E (eds.): Brain Structure, Learning and Memory (ASS Symposium Series). 1986

....can be equal to its performance measure. Furthermore the architectures of the networks can be learned because both the GA and GP can handle highly non linear problems. Unsupervised learning makes no use of correct output patterns or performance measures to allow the adaptation of the network [35, 36, 9, 10, 22]. Instead, the networks usually identify clusters in the input pattern space by assigning a cluster to an output neuron, whereby the output neuron which responds most strongly to the input pattern provides the cluster it belongs to. The learning algorithm is carefully constructed to enable it to ....

....the rule) having class 11. The TRUE child specifies wall pushing should be used in case there is a neuron already to the east of the initial neuron, but since no neuron is there this is ignored. Finally, rule 13 is the only CHANGE ACTIVATION rule present in the individual. It changes 73 1. AC (CXT[1, 2, 3, 4, 6, 7, 10, 14, 15]) 0, 4, 5, 6, 8, 10, 14, 15] FALSE (input) 2. AC (CXT[0, 1, 2, 4, 8, 9, 10, 14, 15] 1, 3, 5, 6, 7, 9, 10, 11, 12, 13, 15, 16] TRUE (output) 3. AC (CXT[2, 3, 4, 9, 10, 13, 14, 16, 17] 0, 2, 3, 5, 6, 8, 9, 10, 13, 14, 16, 17] TRUE (output) 4. AC (CXT[0, 3, 5, 6, 7, 11, 12, 16, 17] 0, 1, ....

[Article contains additional citation context not shown here]

G. A. Carpenter and S. Grossberg. Neural dynamics of category learning and recognition: Attention, memory consolidation, and amnesia. In Joel L. Davis, Robert W. Newburgh, and Edward J. Wegman, editors, Brain Structure, Learning, and Memory, pages 233--290. Boulder, CO: Westview Press, 1998.

....5. 4 Adaptive Resonance Theory The attempt to understand how a learning system can be designed to remain adaptive in response to new significant events, and yet remain stable in response to irrelevant events, has motivated Grossberg and Carpenter in the development of the Adaptive Resonance Theory [28, 11, 12], or ART for short. The above problem, named the stability plasticity dilemma, asks for how can a system preserve its previously learned knowledge while continuing to learn new things, and for what prevents the new knowledge from erasing previously learned things. Moreover, it asks for the ....

....the system to switch in between the two modes, i.e. how the systems decides when to learn from new events, and when to ignore irrelevant information. The Adaptive Resonance Theory introduces the framework under which systems that embody solutions to the above dilemma have been constructed [28, 29, 11, 12, 15, 14]. ART models employ a competitive learning paradigm similar to Kohonen s feature maps, but have added mechanisms for controlling the orientation and vigilance of the learning process. The models are described by a system of parameterized differential equations [11] whose study is beyond the ....

Gail A. Carpenter and Stephen Grossberg. Neural dynamics of category learning and recognition: Attention, memory consolidation, and amnesia. In J. Davis, R. Newburgh, and E. Wegman (Eds.), editor, Brain Structure, Learning, and Memory. AAA Symposium Series, 1987.

....Newell (Newell, 1990) refers to this level as the symbol manipulation level. Thus, neural networks belong to the lowest of these levels. In the last years there has been a growing optimism about the capability of neural networks, both as cognitive models (e.g. the works of Grossberg (Carpenter and Grossberg, 1986) and of Edelman (Edelman, 1989) and as tools for pattern recognition (e.g. backpropagation networks (Rumelhart et al. 1986) However, one must keep in mind that neural networks that can be simulated on a computer, as most neural networks can, are of course at the most Turing machine equivalent. ....

Carpenter, G. and Grossberg, S. (1986). Neural dynamics of category learning and recognition: Attention, memory consolidation, and amnesia. In Davis, J., Newburgh, R., and Wegman, E., editors, Brain Structure, Learning, and Memory, pages 239--286. AAAS Symposium Series.

....Fuzzy ARTMAP network, then we describe, evaluate and discuss an experiment performed on data collected on a real mobile robot, then we conclude and point out possible future research directions. 2. FUZZY ARTMAP ART is a family of neural network structures, based on the Adaptive Resonance Theory [2]. They perform nonlinear, incremental, unsupervised clustering of input data sets. The basic ART module consists of two layers of artificial neurons, interconnected through links of adaptive strengths. Competitive activation rule provides that just a single neuron of the second layer may be active ....

Carpenter,G.A., Grossberg,S. "Neural Dynamics of Category Learning and Recognition: Attention, Memory Consolidation and Amnesia. ", in Davis,J., Newburgh,R. and Wegman,E. (Eds.) Brain Structure, Learning, and Memory, AAAS Symposium Series, 1986.

....method is fine positioning for a door passing task purposes. The tool we implemented is the Fuzzy ARTMAP neural network, described in the following section. 2 Fuzzy ARTMAP for learning localization associations ART is a class of neural network structures, based on the Adaptive Resonance Theory [2], which perform nonlinear, incremental, unsupervised categorization of a set of input data. The fundamental ART algorithm divides input vectors into a number of clusters, in such a way that the data belonging to each cluster are all similar within a specified tolerance. The basic ART module is ....

Carpenter,G.A., Grossberg,S. "Neural Dynamics of Category Learning and Recognition: Attention, Memory Consolidation and Amnesia.", in Davis,J., Newburgh,R. and Wegman,E. (Eds.) Brain Structure, Learning, and Memory, AAAS Symposium Series, 1986.

No context found.

G.A. Carpenter and S. Grossberg. Neural dynamics of category learning and recognition: Attention, memory consolidation, and amnesia. In J. Davis, R. Newburgh, and E. Wegman, editors, Brain Structure, Learning, and Memory, AAAS Symposium Series, pages 239--286, 1986.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC