Neural networks: algorithms, applications, and programming techniques
Neural networks: algorithms, applications, and programming techniques
Medical diagnosis and treatment plans derived from a hybrid expert system
Hybrid architectures for intelligent systems
Representing expert knowledge in neural nets
Hybrid architectures for intelligent systems
Knowledge-based artificial neural networks
Artificial Intelligence
Enhancements to the data mining process
Enhancements to the data mining process
Optimizations of the Combinatorial Neural Model
SBRN '98 Proceedings of the Vth Brazilian Symposium on Neural Networks
Learning in the combinatorial neural model
IEEE Transactions on Neural Networks
Scalable Model for Extensional and Intensional Descriptions of Unclassified Data
IPDPS '00 Proceedings of the 15 IPDPS 2000 Workshops on Parallel and Distributed Processing
A reasoning-based strategy for exploring the synergy among alternative crops
KES'10 Proceedings of the 14th international conference on Knowledge-based and intelligent information and engineering systems: Part III
Characterizing human features: An applied study on proactivity perception of undergraduate students
Pattern Recognition and Image Analysis
Human features recognition with CNM: An applied study concerning undergraduate students
Pattern Recognition and Image Analysis
| Hi-index | 0.00 |
The Combinatorial Neural Model (CNM) ([8] and [9]) is a hybrid architecture for intelligent systems that integrates symbolic and connectionist computational paradigms. This model has shown to be a good alternative to be used on data mining; in this sense some works have been presented in order to deal with scalability of the core algorithm to large databases ([2,1] and [10]). Another important issue is the prunning of the network, after the trainingp hase. In the original proposal this prunningi s done on the basis of accumulators values. However, this criterion does not give a precise notion of the classification accuracy that results after the prunning. In this paper we present an implementation of the CNM with a feature based on the wrapper method ([6] and [12]) to prune the network by usingt he accuracy level, instead of the value of accumulators as in the original approach.