No abstract

No abstract

No abstract

Automatic context recognition enables mobile devices to react to changes in the environment and different situations. While many different sensors can be used for context recognition, the use of acoustic cues is among the most popular and successful. Current approaches to acoustic context recognition (ACR) are too costly in terms of computation and memory requirements to support an always-listening mode. This paper describes our work to develop a reduced complexity, efficient approach to ACR involving support vector machine classifiers. The principal hypothesis is that a significant fraction of training data contains information redundant to classification. Through clustering, training data can thus be selectively decimated in order to reduce the number of support vectors needed to represent discriminative hyperplanes. This represents a significant saving in terms of computational and memory efficiency, with only modest degradations in classification accuracy.

In this paper, we address the problem of determining the equilibrium point of a feedback structure for two-channel adaptive noise cancellation in the presence of crosstalk. We focus on an important characteristic of adaptive filters, namely the steady-state mean-square error that remains after the algorithm has converged but independently of the considered algorithm. Our approach relies on analysis of the relationships between the desired signals and their artifacts (distortion, residual noise) at the system outputs. We show that the equilibrium state is obtained when the energy of the distortion on the output signals is the same on each channel. Using this equilibrium state, we provide answers to questions for which no satisfactory answers are currently available for this structure. Examples are given to illustrate that even qualitatively, these answers can be good approximations. Results of simulations sustain our claims.