What question did this study set out to answer?

The aim is to develop a new loss function, MPCE, to enhance neural network classification accuracy.

January 1, 2019IEEE AccessOpen Access

MPCE: A Maximum Probability Based Cross Entropy Loss Function for Neural Network Classification

View Full Paper

Ask AI

Bookmark

Authors

Randomized trial demonstrates improved accuracy in neural network classification, indicating better learning efficiency.

The aim is to develop a new loss function, MPCE, to enhance neural network classification accuracy.
Developed a novel cross entropy loss function called MPCE based on maximum probability.
Analyzed gradient differences between MPCE and standard cross entropy loss.
Conducted experiments across six public datasets to evaluate the proposed algorithm's performance.
MPCE shows faster convergence compared to other loss functions.
Achieved the highest training and test accuracy on six datasets.
Outperformed other methods in classification accuracy, especially on the sensor dataset.

Zhou et al. (2019) studied this question.

View Full Paper

Ask AI

Bookmark