Introduction - Hierarchical Neural Networks for Image Interpretation - page 14

Supervised learning is also applicable to the architecture. It is proposed to com-
bine the RPROP optimization method with backpropagation through time to achieve
stable and fast learning. This supervised training is applied to several learning tasks.
A feed-forward block classifier is trained to recognize meter values without the
need for prior digit segmentation. It is combined with a digit classifier if necessary.
The system is able to recognize meter values that are challenging for human experts.
A recurrent network is trained to binarize matrix codes. The desired outputs are
produced by applying an adaptive thresholding method to undegraded images. The
network is trained to produce the same output even for images that have been de-
graded with typical noise. It learns to recognize the cell structure of the matrix codes.
The binarization performance is evaluated using a recognition system. The trained
network performs better than the adaptive thresholding method for the undegraded
images and outperforms it significantly for degraded images.
The architecture is also applied for the learning of image reconstruction tasks.
Images are degraded and networks are trained to reproduce the originals iteratively.
For a super-resolution problem, small recurrent networks are shown to outperform
feed-forward networks of similar complexity. A larger network is used for the
filling-in of occlusions, the removal of noise, and the enhancement of image con-
trast. The network is also trained to reconstruct images from a sequence of degraded
images. It is able to solve this task even in the presence of high noise.
Finally, the proposed architecture is applied for the task of face localization.
A recurrent network is trained to localize faces of different individuals in complex
office environments. This task is challenging due to the high variability of the dataset
used. The trained network performed significantly better than the hybrid localization
method, proposed by the creators of the dataset. It is not limited to static images,
but can track a moving face in real time.