Abstract | ||
---|---|---|
•Determination of CNN parameters with an interpretable feedforward design methodology.•Development of the subspace approximation with adjusted bias (Saab) signal transform.•Design of fully-connected layers as a cascade of linear least squared regressors.•Comparison of CNNs designed by backpropagation and feedforward methodologies. |
Year | DOI | Venue |
---|---|---|
2018 | 10.1016/j.jvcir.2019.03.010 | Journal of Visual Communication and Image Representation |
Keywords | Field | DocType |
Interpretable machine learning,Convolutional neural networks,Principal component analysis,Linear least-squared regression,Cross entropy,Dimension reduction | MNIST database,Subspace topology,Pattern recognition,Convolutional neural network,Robustness (computer science),Cascade,Artificial intelligence,Backpropagation,Mathematics,Principal component analysis,Feed forward | Journal |
Volume | ISSN | Citations |
60 | 1047-3203 | 12 |
PageRank | References | Authors |
0.56 | 13 | 5 |
Name | Order | Citations | PageRank |
---|---|---|---|
C.-C. Jay Kuo | 1 | 7524 | 697.44 |
Min Zhang | 2 | 12 | 0.56 |
Siyang Li | 3 | 29 | 4.55 |
Jiali Duan | 4 | 13 | 2.94 |
Yueru Chen | 5 | 19 | 3.50 |