Connectionist learning procedures
Artificial Intelligence
Methods for combining experts' probability assessments
Neural Computation
Smart feature detection using an invariance network architecture
BMVC '95 Proceedings of the 1995 British conference on Machine vision (Vol. 1)
Neural Computation
Hi-index | 0.00 |
In this paper, we study a bootstrapped learning procedure applied to corner detection using synthetic training data generated from a grey-level model of a corner feature which permits sampling of the pattern space at arbitrary density as well as providing a self-consistent validation set to assess the classifier generalisation. Since adequate learning of the whole mapping by a single neural network is problematic we partition data across modules using bootstrapping and which we then combine by a meta-learning stage. We test the hierarchical classifier on real images and compare results with those obtained by a monolithic network.