If the dataset available to machine learning results from cluster sampling (e.g. patients from a sample of hospital wards), the usual cross-validation error rate estimate can lead...
In the current information age, the dominant method for information search is by providing few keywords to a search engine. Keyword search is currently one of the most important o...
Self-Organizing Maps (SOM) are very powerful tools for data mining, in particular for visualizing the distribution of the data in very highdimensional data sets. Moreover, the 2D m...
The complexity of numerical domain partitioning depends on the number of potential cut points. In multiway partitioning this dependency is often quadratic, even exponential. There...
The intelligent reformulation or restructuring of a belief network can greatly increase the efficiency of inference. However, time expended for reformulation is not available for ...