The nature of statistical learning theory
The nature of statistical learning theory
Three learning phases for radial-basis-function networks
Neural Networks
MindReader: Querying Databases Through Multiple Examples
VLDB '98 Proceedings of the 24rd International Conference on Very Large Data Bases
Comparing support vector machines with Gaussian kernels to radialbasis function classifiers
IEEE Transactions on Signal Processing
IEEE Transactions on Image Processing
Relevance feedback: a power tool for interactive content-based image retrieval
IEEE Transactions on Circuits and Systems for Video Technology
A geometrical representation of McCulloch-Pitts neural model and its applications
IEEE Transactions on Neural Networks
Interactive Image Retrieval in a Fuzzy Framework
WILF '09 Proceedings of the 8th International Workshop on Fuzzy Logic and Applications
The state of the art in image and video retrieval
CIVR'03 Proceedings of the 2nd international conference on Image and video retrieval
Image retrieval using fuzzy relevance feedback and validation with MPEG-7 content descriptors
PReMI'07 Proceedings of the 2nd international conference on Pattern recognition and machine intelligence
Hi-index | 0.00 |
It has been generally acknowledged that relevance feedback in image retrieval can be considered as a two-class learning and classification process. The classifier used is essential to the performance of relevance feedback. In this paper, a RBF neural network is employed during the relevance feedback process. The architecture of the RBF network is automatically determined by the constructive learning algorithm (CLA). The weights in the output layer of the network are learned by Least-mean-square method. Experiment results on 10,000 heterogeneous images demonstrate the proposed CLA-based RBF network can achieve comparable performance with support vector machines and support vector learning based RBF during the relevance feedback process. Furthermore, a practical advantage of the CLA-based RBF network is that the width of Gaussian kernel does not need to manually set while for SVM it need to be predefined according to experience.