Topic
Maximum a posteriori estimation
About: Maximum a posteriori estimation is a research topic. Over the lifetime, 7486 publications have been published within this topic receiving 222291 citations. The topic is also known as: Maximum a posteriori, MAP & maximum a posteriori probability.
Papers published on a yearly basis
Papers
More filters
TL;DR: Applications to well‐known problems of distribution fitting, quantal responses and least‐squares curve fitting, and sequential minimization and nested minimization can be used to solve particular problems are described.
Abstract: Maximum‐likelihood estimation problems can be solved numerically using function minimization algorithms, but the amount of computing required and the accuracy of the results depend on the way the algorithms are used. Attention to the analytical properties of the model, to the relationship between the model and the data, and to descriptive properties of the data can greatly simplify the problem, sometimes providing a method of solution on a desk calculator. This paper describes how parameter transformation, sequential minimization and nested minimization can be used to solve particular problems. Applications to well‐known problems of distribution fitting, quantal responses and least‐squares curve fitting are described. The implications for computer programming are discussed.
108 citations
TL;DR: Two classifier approaches, namely classifiers based on Multi Layer Perceptrons (MLPs) and Gaussian Mixture Models (GMMs), are compared for use in a face verification system and it is shown that for low resolution faces the MLP approach has slightly lower error rates than theGMM approach; however, the GMM approach easily outperforms theMLP approach for high resolution faces and is significantly more robust to imperfectly located faces.
Abstract: We compare two classifier approaches, namely classifiers based on Multi Layer Perceptrons (MLPs) and Gaussian Mixture Models (GMMs), for use in a face verification system. The comparison is carried out in terms of performance, robustness and practicability. Apart from structural differences, the two approaches use different training criteria; the MLP approach uses a discriminative criterion, while the GMM approach uses a combination of Maximum Likelihood (ML) and Maximum a Posteriori (MAP) criteria. Experiments on the XM2VTS database show that for low resolution faces the MLP approach has slightly lower error rates than the GMM approach; however, the GMM approach easily outperforms the MLP approach for high resolution faces and is significantly more robust to imperfectly located faces. The experiments also show that the computational requirements of the GMM approach can be significantly smaller than the MLP approach at a cost of small loss of performance.
108 citations
TL;DR: A new method for clustering functional data is proposed under the name Funclust, which relies on the approximation of the notion of probability density for functional random variables, which generally does not exist and a parametric mixture model is proposed.
108 citations
TL;DR: A multitarget detection/tracking (D/T) algorithm is proposed, which exploits the lateral continuity of echoes arising from a multilayered medium to make layer stripping useful.
Abstract: Monostatic ground penetrating radar (GPR) has proven to be a useful technique in pavement profiling. In road and highway pavements, layer thickness and permittivity of asphalt and concrete can be estimated by using an inverse scattering approach. Layer-stripping inversion refers to the iterative estimation of layer properties from amplitude and time of delay (TOD) of echoes after their detection. This method is attractive for real-time implementation, in that accuracy is improved by reducing false alarms. To make layer stripping useful, a multitarget detection/tracking (D/T) algorithm is proposed. It exploits the lateral continuity of echoes arising from a multilayered medium. Interface D/T means that both detection and tracking are employed simultaneously (not sequentially). For each scan, both detection of the target and tracking of the corresponding TOD of the backscattered echoes are based on the evaluated a posteriori probability density. The TOD is then estimated by using the maximum a posteriori (MAP) or the minimum mean square error (MMSE) criterion. The statistical properties of a scan are related to those of the neighboring ones by assuming, for the interface, a first-order Markov model.
108 citations
TL;DR: It is shown that more accurate and robust results may be obtained through seeking a joint solution to these linked processes through applying Markov random fields in the solution of a maximum a posteriori model of segmentation and registration.
108 citations