Abstract Details
107 / 2023-09-19 19:41:51
An Evolutionary Algorithm-based Neural Architecture Search for Feature Separation in Classification Problems
convolutional neural network,neural architecture search,evolutionary algorithm,feature separation
Final Paper
在本文中，我们提出了基于进化算法的神经架构搜索（EvoNAS），利用进化算法寻找优化的卷积神经网络（CNN），并结合分布估计来提高搜索过程中的效率，从而提高分类性能。优化后的 CNN 揭示了分类过程。它可以将特征按照指定的类别进行分类，在达到期望的分类精度的同时提高相似特征的分类精度。我们进行特征分离实验在公开的CIFAR数据集上，我们使用EvoNAS进行特征分离实验。EvoNAS 在公开可用的 CIFAR 数据集上进行了特征分离实验。分类准确率分别提高了0.55%和0. 与原网络相比分别提高了56\%。此外，特征可视化和归因可视化提高了CNN的分类可解释性。
Important Date
-
Conference Date
Nov 02
2023
toNov 04
2023
-
Dec 15 2023
Draft paper submission deadline
-
Dec 20 2023
Registration deadline
Sponsored By
IEEE Instrumentation and Measurement Society
Xidian University
Xidian University
Contact Information
Previous Conferences
-
2025-11-21 China Guangzhou
2025 International Conference on Sensing, Measurement & Data Analytics in the era of Artificial Intelligence -
2023-11-02 China Xi'an
2023 International Conference on Sensing, Measurement & Data Analytics in the era of Artificial Intelligence -
2022-11-30 China Harbin
International Conference on Sensing, Measurement & Data Analytics in the era of Artificial Intelligence -
2021-10-21 China Nanjing
2021 International Conference on Sensing, Measurement and Data Analytics in the era of Artificial Intelligence -
2020-10-15 China Xi'an
International Conference on Sensing, Measurement and Data Analytics in the era of Artificial Intelligence
