On the Problem of Local Minima in Backpropagation
IEEE Transactions on Pattern Analysis and Machine Intelligence
Integrating neural networks with image pyramids to learn target context
Neural Networks - Special issue: automatic target recognition
Learning from Examples with Information Theoretic Criteria
Journal of VLSI Signal Processing Systems
Clustering and Classification in Structured Data Domains Using Fuzzy Lattice Neurocomputing (FLN)
IEEE Transactions on Knowledge and Data Engineering
Speech Synthesis Using Neural Networks Trained by an Evolutionary Algorithm
ICCS '01 Proceedings of the International Conference on Computational Science-Part II
Adaptive Graphical Pattern Recognition Beyond Connectionist-Based Approaches
Proceedings of the Joint IAPR International Workshops on Advances in Pattern Recognition
Object Detection Using the Statistics of Parts
International Journal of Computer Vision
Artificial Neural Networks for Document Analysis and Recognition
IEEE Transactions on Pattern Analysis and Machine Intelligence
A Toplogy Based Multi-Classifier System
ICDAR '05 Proceedings of the Eighth International Conference on Document Analysis and Recognition
Random Embedding Machines for Pattern Recognition
Neural Computation
Filtering segmentation cuts for digit string recognition
Pattern Recognition
Using Gaussian Processes in Bayesian Robot Programming
IWANN '09 Proceedings of the 10th International Work-Conference on Artificial Neural Networks: Part II: Distributed Computing, Artificial Intelligence, Bioinformatics, Soft Computing, and Ambient Assisted Living
Neural networks for fitting PES data distributions of asphaltene interaction
IJCNN'09 Proceedings of the 2009 international joint conference on Neural Networks
Cross-lingual English Spanish tonal accent labeling using decision trees and neural networks
NOLISP'11 Proceedings of the 5th international conference on Advances in nonlinear speech processing
Protein fold recognition with combined SVM-RDA classifier
HAIS'10 Proceedings of the 5th international conference on Hybrid Artificial Intelligence Systems - Volume Part I
Learning with boundary conditions
Neural Computation
| Hi-index | 0.14 |
This paper discusses the ability of multilayer perceptrons (MLPs) to model the probability distribution of data in typical pattern recognition and verification problems. It is proven that multilayer perceptrons with sigmoidal units and a number of hidden units less or equal than the number of inputs are unable to model patterns distributed in typical clusters, since these networks draw open separation surfaces in the pattern space. When using more hidden units than inputs, the separation surfaces can be closed but, unfortunately, it is proven that determining whether or not an MLP draws closed separation surfaces in the pattern space is ${\cal NP}$-hard. The major conclusion of this paper is somewhat opposite to what is believed and reported in many application papers: MLPs are definitely not adequate for applications of pattern recognition requiring a reliable rejection and, especially, they are not adequate for pattern verification tasks.