We study in this paper the consequences of using the Mean Absolute Percentage Error (MAPE) as a measure of quality for regression models. We show that finding the best model under the MAPE is equivalent to doing weighted Mean Absolute Error (MAE) regression. We show that universal consistency of Empirical Risk Minimization remains possible using the MAPE instead of the MAE.

A new approach is proposed for finding optimal cuts in hierarchies of partitions by energy minimization. It rests on the notion of h-increasingness, allows to find best(optimal) cuts in one pass, and to obtain nice ”climbing” scale space operators. The ways to construct h-increasing energies, and to combine them are studied, and illustrated by two examples on color and on textures.

We propose a new method to precisely detect pupil contours in faces. The method is based on minimizing the energy of pattern and edge. The basic idea of this method is that the energy, which consists of the pattern and the edge energy, has to be minimized. An efficient search method is also introduced to overcome the underlying problem of efficiency of energy minimization methods. "Guide patter...

This paper proposes a system to depth order regions of a frame belonging to a monocular image sequence. For a given frame, regions are ordered according to their relative depth using the previous and following frames. The algorithm estimates occluded and disoccluded pixels belonging to the central frame. Afterwards, a Binary Partition Tree (BPT) is constructed to obtain a hierarchical, region b...

An EPG-representation of a graph G is a collection of paths in a plane square grid, each corresponding to a single vertex of G, so that two vertices are adjacent if and only if their corresponding paths share infinitely many points. In this paper we focus on graphs admitting EPG-representations by paths with at most 2 bends. We show hardness of the recognition problem for this class of graphs, ...

We have designed a simple method to place particles on lattices, concentric shells and icosahedral concentric layers for minimizing the total energy of Lennard-Jones clusters, approximately, by analytical means. The most signiicant diierence of our schemes from others is the dramatic reduction of parameters, which allows the study of large clusters, not possible otherwise. We present the deriva...

We report on some extensions of the Kandinsky model: A new and highly nontrivial technique to incorporate nonplanar drawings into the Kandinsky model in the same way as in the GIOTTO approach is presented. This means a major step towards the practical usability of our approach. The used technique even gives new insights for the solvability of network ow problems. Another variant of Kandinsky en...

This paper introduces a class of resonant controllers that can be used to minimize structural vibration using collocated piezoelectric actuator-sensor pairs. The proposed controller increases the damping of the structure so as to minimize a chosen number of resonant responses. The controller can be tuned to a chosen number of modes. This results in controllers of minimal dimension. The controll...

Support Vector Machine (SVM) is a classification technique based on Structural Risk Minimization (SRM), which can run on MATLAB. For classification of nonseparable samples, conventional SVM needs to select a tradeoff between maximization the margin and misclassification rate. In order to guarantee generalized performance and low misclassification rate of SVM, this paper puts forward an improved...