Figure - PMC

Skip to main content

An official website of the United States government

Here's how you know

Here's how you know

Official websites use .gov
A .gov website belongs to an official government organization in the United States.

Secure .gov websites use HTTPS
A lock ( ) or https:// means you've safely connected to the .gov website. Share sensitive information only on official, secure websites.

View full-text article in PMC

. 2020 Jul 31;3:54. doi: 10.3389/frai.2020.00054

Search in PMC
Search in PubMed
View in NLM Catalog
Add to search

Copyright © 2020 Cheng, Nazarian and Bogdan.

This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

PMC Copyright notice

Trustworthiness comparison among three training strategies. (A) After one epoch of sample by sample training, the accuracy is 20.11%. The neural network is underfitting and stable, i.e., the loss is not decreasing in the first epoch of sample by sample training. The trustworthiness results from “pre-train” and “evolution” versions are the same. (B) After one epoch of mini-batch training, the accuracy is 47.23%. The neural network is underfitting and very unstable. Evaluating the trustworthiness on more data samples minimizes the gap between “pre-train” and “evolution” versions. (C) After multiple epochs of mini-batch training the accuracy achieves 59.49%. The neural network is semi-stable and the trustworthiness evaluation results between the two versions get closer if more data samples are used for trustworthiness evaluation.