☆ 4.7 Letter

Constructing an automatic diagnosis and severity-classification model for acromegaly using facial photographs by deep learning

JOURNAL OF HEMATOLOGY & ONCOLOGY (2020)

Bias in Cross-Entropy-Based Training of Deep Survival Networks

Shekoufeh Gorgi Zadeh et al.

Summary: Research shows that training deep survival networks with the cross-entropy loss function may result in high prediction error and bias. In contrast, using the negative log-likelihood loss function leads to better calibrated prediction rules and improved discriminatory power.

IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE (2021)

添加到收藏夹

Article Chemistry, Analytical