Publica
Hier finden Sie wissenschaftliche Publikationen aus den FraunhoferInstituten. Towards optimal nonlinearities for sparse recovery using higherorder statistics
 Palmieri, F.A.N. ; Institute of Electrical and Electronics Engineers IEEE; IEEE Signal Processing Society: MLSP 2016, IEEE International Workshop on Machine Learning for Signal Processing. Proceedings : September 1316, Vietri sul Mare, Salerno, Italy Piscataway, NJ: IEEE, 2016 ISBN: 9781509007462 ISBN: 9781509007479 S.483488 
 International Workshop on Machine Learning for Signal Processing (MLSP) <26, 2016, Salerno> 

 Englisch 
 Konferenzbeitrag 
 Fraunhofer HHI () 
Abstract
We consider machine learning techniques to develop lowlatency approximate solutions for a class of inverse problems. More precisely, we use a probabilistic approach to the problem of recovering sparse stochastic signals that are members of the lpballs. In this context, we analyze the Bayesian meansquareerror (MSE) for two types of estimators: (i) a linear estimator and (ii) a structured estimator composed of a linear operator followed by a Cartesian product of univariate nonlinear mappings. By construction, the complexity of the proposed nonlinear estimator is comparable to that of its linear counterpart since the nonlinear mapping can be implemented efficiently in hardware by means of lookup tables (LUTs). The proposed structure lends itself to neural networks and iterative shrinkage/thresholdingtype algorithms restricted to a single iteration (e.g. due to imposed hardware or latency constraints). By resorting to an alternating minimization technique, we obtain a sequence of optimized linear operators and nonlinear mappings that converge in the MSE objective. The result is attractive for realtime applications where general iterative and convex optimization methods are infeasible.