In this paper, we propose a regularization technique for AdaBoost. The method implements a bias-variance control strategy in order to avoid overfitting in classification tasks on noisy data. The method is based on a notion of easy and hard training patterns as emerging from analysis of the dynamical evolutions of AdaBoost weights. The procedure consists in sorting the training data points by a hardness measure, and in progressively eliminating the hardest, stopping at an automatically selected threshold. Effectiveness of the method is tested and discussed on synthetic as well as real data.

The identifiers chosen by programmers as entity names contain valuable information. They are often the starting point for the program understanding activities, especially when high level views, like the call graph, are available. An approach for the restructuring of program identifier names is proposed, aimed at improving their meaningfulness. It considers two forms of standardization, associated respectively to the lexicon of the composing terms and to the syntax of their arrangement. Automatic and semiautomatic techniques are described which can help the restructuring intervention. Their application to a real world case study is also presented.

This paper deals with the development of a navigation system for a mobile robot designed to accomplish missions inside the building of big organisations, such as hospitals or malls. Useful terminology is first introduced, and followed by a rather detailed presentation of the architecture and overall functioning of the system. The most recent realization of the system is then described in some detail, and its behaviours and functionalities are reported and discussed. Possible directions for future work are outlined, along with open problems that are regarded as the most interesting and challenging for the further development of the system.>

Two aspects of edge detection are analyzed, namely accuracy of localization and sensitivity to noise. The detection of corners and trihedral vertices is analyzed for gradient schemes and zero-crossing schemes. It is shown that neither scheme correctly detects corners of trihedral vertices, but that the gradient schemes are less sensitive to noise. A simple but important conclusion is that the noise present in digital images of typical indoor scenes is small and the signal-to-noise ratio is high. The noise present in digital images is so small as to make the performances of a variety of filters almost indistinguishable. As a consequence small filters can be used and the exact shape of the filter is not critical.>