Asymptotic performance loss in bayesian hypothesis testing under data quantization
S Jana - 2009 43rd Annual Conference on Information …, 2009 - ieeexplore.ieee.org
2009 43rd Annual Conference on Information Sciences and Systems, 2009•ieeexplore.ieee.org
In a variety of decision systems, processing is performed not on the underlying signal but on
a quantized version. Accordingly, assuming fine quantization, Poor observed a quadratic
variation in f-divergences with smooth f. In contrast, we derive a quadratic behavior in the
Bayesian probability of error, which corresponds to a nonsmooth f, thereby advancing the
state of the art. Unlike Poor's purely variational method, we solve a novel cube-slicing
problem, and convert a volume integral to a surface integral in the course of our analysis. In …
a quantized version. Accordingly, assuming fine quantization, Poor observed a quadratic
variation in f-divergences with smooth f. In contrast, we derive a quadratic behavior in the
Bayesian probability of error, which corresponds to a nonsmooth f, thereby advancing the
state of the art. Unlike Poor's purely variational method, we solve a novel cube-slicing
problem, and convert a volume integral to a surface integral in the course of our analysis. In …
In a variety of decision systems, processing is performed not on the underlying signal but on a quantized version. Accordingly, assuming fine quantization, Poor observed a quadratic variation in f-divergences with smooth f. In contrast, we derive a quadratic behavior in the Bayesian probability of error, which corresponds to a nonsmooth f, thereby advancing the state of the art. Unlike Poor's purely variational method, we solve a novel cube-slicing problem, and convert a volume integral to a surface integral in the course of our analysis. In this paper, we elaborate our method, and sharpen our result, a preliminary version of which were outlined in our previous work.
ieeexplore.ieee.org
顯示最佳搜尋結果。 查看所有結果