In this paper' two different approaches to non-linearity simplification in neural nets are presented. Both the solutions are based on approximation of the sigmoidal mapper often used in neural networks (extensions are being considered to allow approximation of a more general class of functions). In particular a first solution yielding a very simple architecture, but involving discontinuous functions is presented; a second solution, slightly more complex, but based on a continuous function is then presented. This second solution has been successfully used in conjunction with the classical generalized delta rule algorithm. Considering the negative semi-axis and two consecutive points P,, and P,-l (as shown in figure 3) where.

No abstract

SUMMARYPower electronic elements are often built employing arrays of devices designed to evenly share high currents over a large number of single devices. In some situations, mismatch in current sharing can occur causing temperature differences between elements of the array. These differences can trigger electrothermal instability phenomena and even device destruction. In this paper, we consider the electrothermal dynamic behavior of arrays composed of identical one-port elements. We first introduce a simplified model of the thermal paths and of the elements composing the array that allows the analysis of the behavior of the electrothermal system. Then, we investigate from a mathematical standpoint dynamic effects governing current sharing and possible situations causing hot spots with consequent device destruction. These results are compared with simulations of detailed non-linear models confirming predictions given by the simplified model.

No abstract

No abstract