Journal
IEEE TRANSACTIONS ON CYBERNETICS
Volume 52, Issue 5, Pages 3221-3231Publisher
IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
DOI: 10.1109/TCYB.2020.3009582
Keywords
Optimization; Training; Adaptive learning; Evolutionary computation; Artificial neural networks; Decoding; Adaptive learning framework; deep neural network (DNN); evolutionary algorithm; multiobjective optimization algorithm; weighted sparseness
Categories
Funding
- National Natural Science Foundation of China [61701238, 11431015, 61773209, 61873148, 61933007]
- Natural Science Foundation of Jiangsu Province of China [BK20190021]
- Six Talent Peaks Project in Jiangsu Province of China [XYDXX-033]
- Alexander von Humboldt Foundation of Germany
Ask authors/readers for more resources
In this article, an adaptive learning framework for a deep weighted sparse autoencoder is established using the multiobjective evolutionary algorithm. The framework successfully optimizes the autoencoder by introducing weighted sparsity to impose varying degrees of sparse constraints on the hidden units. Experimental results demonstrate the effectiveness and generality of the framework, which is also applied to image quality assessment.
In this article, an adaptive learning framework is established for a deep weighted sparse autoencoder (AE) by resorting to the multiobjective evolutionary algorithm (MOEA). The weighted sparsity is introduced to facilitate the design of the varying degrees of the sparsity constraints imposed on the hidden units of the AE. The MOEA is exploited to adaptively seek appropriate hyperparameters, where the divide-and-conquer strategy is implemented to enhance the MOEA's performance in the context of deep neural networks. Moreover, a sharing scheme is proposed to further reduce the time complexity of the learning process at the slight expense of the learning precision. It is shown via extensive experiments that the established adaptive learning framework is effective, where different sparse models are utilized to demonstrate the generality of the proposed results. Then, the generality of the proposed framework is examined on the convolutional AE and VGG-16 network. Finally, the developed framework is applied to the blind image quantity assessment that illustrates the applicability of the established algorithms.
Authors
I am an author on this paper
Click your name to claim this paper and add it to your profile.
Reviews
Recommended
No Data Available