On estimation and prediction of geostatistical regression models via a corrected Stein's unbiased risk estimator

2016 ◽  
Vol 28 (1) ◽  
pp. e2424 ◽  
Author(s):  
Hong-Ding Yang ◽  
Chun-Shu Chen
Keyword(s):  
Regression Models ◽  
Estimation And Prediction ◽  
Risk Estimator ◽  
Unbiased Risk

scholarly journals Image Denoising Based on Nonlocal Bayesian Singular Value Thresholding and Stein’s Unbiased Risk Estimator

2019 ◽  
Vol 28 (10) ◽  
pp. 4899-4911 ◽  
Author(s):  
Caoyuan Li ◽  
Hong-Bo Xie ◽  
Xuhui Fan ◽  
Richard Yi Da Xu ◽  
Sabine Van Huffel ◽  
...  
Keyword(s):  
Image Denoising ◽  
Singular Value ◽  
Risk Estimator ◽  
Unbiased Risk ◽  
Singular Value Thresholding

scholarly journals Learning from Complementary Labels via Partial-Output Consistency Regularization

2021 ◽  
Author(s):  
Deng-Bao Wang ◽  
Lei Feng ◽  
Min-Ling Zhang
Keyword(s):  
Weighting Factor ◽  
Training Phase ◽  
Competitive Performance ◽  
Regularization Techniques ◽  
Series Of Experiments ◽  
Risk Estimator ◽  
Dynamic Weighting ◽  
Unbiased Risk

In complementary-label learning (CLL), a multi-class classifier is learned from training instances each associated with complementary labels, which specify the classes that the instance does not belong to. Previous studies focus on unbiased risk estimator or surrogate loss while neglect the importance of regularization in training phase. In this paper, we give the first attempt to leverage regularization techniques for CLL. By decoupling a label vector into complementary labels and partial unknown labels, we simultaneously inhibit the outputs of complementary labels with a complementary loss and penalize the sensitivity of the classifier on the partial outputs of these unknown classes by consistency regularization. Then we unify the complementary loss and consistency loss together by a specially designed dynamic weighting factor. We conduct a series of experiments showing that the proposed method achieves highly competitive performance in CLL.

Identifying Potential Care Management Candidates: The Use of Diagnosis-Based Models to Predict 5-Year Health Care Costs

2016 ◽  
Vol 17 (2) ◽  
pp. 105-111
Author(s):  
John Robst
Keyword(s):  
Health Care ◽  
Case Management ◽  
Regression Models ◽  
Care Management ◽  
Predictive Power ◽  
Service Use ◽  
Claims Data ◽  
Individual Characteristics ◽  
Estimation And Prediction ◽  
Care Costs

Objective: This article examined individual characteristics associated with having higher costs in a 5-year period to identify patients that may potentially benefit from case management.Methods: Florida Medicaid claims data from 2005 to 2010 were used to examine the characteristics, diagnoses, and services (in 2005) associated with individual costs in 5 future years (2006–2010). The data were divided into estimation and prediction samples with regression models estimated using diagnoses and service use in 2005 to predict future costs. Predictive power was assessed by applying the model results to the prediction sample and comparing predicted costs to actual costs.Results: Demographics, service use, and diagnosis in 2005 were associated with costs in the following 5 years. Models were predictive of future costs with a significant relationship between the predicted costs and actual costs.Conclusion: Diagnosis-based models in conjunction with prior costs can predict future costs. Individuals predicted to have higher costs may be candidates for case management to potentially avoid reduce costs.

IMAGE DENOISING BASED ON WAVELET SHRINKAGE USING NEIGHBOR AND LEVEL DEPENDENCY

2009 ◽  
Vol 07 (03) ◽  
pp. 299-311 ◽  
Author(s):  
DONGWOOK CHO ◽  
TIEN D. BUI ◽  
GUANGYI CHEN
Keyword(s):  
Signal To Noise Ratio ◽  
Signal Denoising ◽  
Wavelet Thresholding ◽  
Wavelet Coefficients ◽  
Wavelet Shrinkage ◽  
Minimum Risk ◽  
Shrinkage Methods ◽  
Risk Estimator ◽  
Visual Artifacts ◽  
Unbiased Risk

Since Donoho et al. proposed the wavelet thresholding method for signal denoising, many different denoising approaches have been suggested. In this paper, we present three different wavelet shrinkage methods, namely NeighShrink, NeighSure and NeighLevel. NeighShrink thresholds the wavelet coefficients based on Donoho's universal threshold and the sum of the squares of all the wavelet coefficients within a neighborhood window. NeighSure adopts Stein's unbiased risk estimator (SURE) instead of the universal threshold of NeighShrink so as to obtain the optimal threshold with minimum risk for each subband. NeighLevel uses parent coefficients in a coarser level as well as neighbors in the same subband. We also apply a multiplying factor for the optimal universal threshold in order to get better denoising results. We found that the value of the constant is about the same for different kinds and sizes of images. Experimental results show that our methods give comparatively higher peak signal to noise ratio (PSNR), are much more efficient and have less visual artifacts compared to other methods.

scholarly journals Estimating the Rank of a Nonnegative Matrix Factorization Model for Automatic Music Transcription Based on Stein’s Unbiased Risk Estimator

2020 ◽  
Vol 10 (8) ◽  
pp. 2911
Author(s):  
Seokjin Lee
Keyword(s):  
Matrix Factorization ◽  
Nonnegative Matrix Factorization ◽  
A Priori ◽  
Nonnegative Matrix ◽  
Music Transcription ◽  
Estimation Algorithms ◽  
Automatic Music Transcription ◽  
Risk Estimator ◽  
Unbiased Risk ◽  
Basis Vectors

In this paper, methods to estimate the number of basis vectors of the nonnegative matrix factorization (NMF) of automatic music transcription (AMT) systems are proposed. Previously, studies on NMF-based AMT have demonstrated that the number of basis vectors affects the performance and that the number of note events can be a good selection as the rank of NMF. However, many NMF-based AMT methods do not provide a method to estimate the appropriate number of basis vectors; instead, the number is assumed to be given in advance, even though the number of basis vectors significantly affects the algorithm’s performance. Recently, based on Bayesian methods, certain estimation algorithms for the number of basis vectors have been proposed; however, they are not designed to be used as music transcription algorithms but are components of specific NMF methods and thus cannot be used generally as NMF-based transcription algorithms. Our proposed estimation algorithms are based on eigenvalue decomposition and Stein’s unbiased risk estimator (SURE). Because the SURE method requires variance in undesired components as a priori knowledge, the proposed algorithms estimate the value using random matrix theory and first and second onset information in the input music signal. Experiments were then conducted for the AMT task using the MIDI-aligned piano sounds (MAPS) database, and these algorithms were compared with variational NMF, gamma process NMF, and NMF with automatic relevance determination algorithms. Based on experimental results, the conventional NMF-based transcription algorithm with the proposed rank estimation algorithms demonstrated enhanced F1 score performances of 2–3% compared to the algorithms. While the performance advantages are not significantly large, the results are meaningful because the proposed algorithms are lightweight, are easy to combine with any other NMF methods that require an a priori rank parameter, and do not have setting parameters that considerably affect the performance.

Sign in / Sign up

Export Citation Format

Share Document