A new handwriting model is proposed for simulating complex handwriting variation in obedience to individual handwriting tendencies. Individual handwriting t,endency is represented...
We demonstrate in this paper how certain forms of rule-based knowledge can be used to prestructure a neural network of normalized basis functions and give a probabilistic interpre...
Bayesian methods are valuable, inter alia, whenever there is a need to extract information from data that is uncertain or subject to any kind of error or noise (including measurem...
In the analysis of a newspaper page an important step is the clustering of various text blocks into logical units, i.e., into articles. We propose three algorithms based on text p...
Tutoring systems are described as having two loops. The outer loop executes once for each task, where a task usually consists of solving a complex, multi-step problem. The inner lo...