We introduce a trainable system that simultaneously filters and classifies low-level features into types specified by the user. The system operates over full colour images, and ou...
For the tasks of classification, two types of patterns can generate problems: ambiguous patterns and outliers. Furthermore, it is possible to separate classification algorithms in...
Jonathan Milgram, Mohamed Cheriet, Robert Sabourin
Global and local adaptive thresholding techniques have been shown effective on particular types of documents. None produces consistently good results on all types of documents. In...
In this paper, we sketch a method for clustering e-commerce search engines by the type of products/services they sell. This method utilizes the special features of interface pages...
We define seal, an untyped call-by-value -calculus with primitives for protecting abstract data by sealing, and develop a bisimulation proof method that is sound and complete with...