Adaptive hyperparameter and regularization strategies for improved elastic weight consolidation in lung cancer classification | International Journal for Simulation and Multidisciplinary Design Optimization (IJSMDO)

AI-Driven Simulation and Neural Optimization for Smart Systems and Healthcare Applic

Open Access

Table 6

Dynamic hyperparameter for Task3.

Input 1000 image (ACA &N& SCC) (800 image for training & 200 for validation) (balanced training samples 627 image)
Epoch	Train loss	Train accuracy	Test loss	Test accuracy	Test precision	Test recall	Test F1-score	EWC penalty	EWC lambda	Learning rate	Time
1	0.3425	89.31%	0.0808	97.5%	98%	97%	98%	743.19	14863.87	0.0002	9s
2	0.2153	93.3%	0.0906	96.5%	97%	96%	96%	701.05	14021.04	0.0002	9s
3	0.1177	94.74%	0.0759	98%	98%	98%	98%	665.42	13308.4	0.0002	9s
4	0.0906	96.17%	0.0669	98%	98%	98%	98%	642.33	12846.6	0.0002	9s
5	0.0695	96.97%	0.0654	98%	98%	98%	98%	620.17	12403.4	0.0002	9s
6	0.0554	97.77%	0.0578	98%	98%	98%	98%	597.41	11948.2	0.0002	9s
7	0.0605	97.13%	0.041	99%	99%	99%	99%	575.11	11502.2	0.0002	9s
8	0.0403	98.56%	0.0499	98%	99%	97%	99%	553.22	11064.4	0.0002	9s
9	0.0631	97.13%	0.0645	97.5%	98%	97%	98%	536.88	10737.6	0.0002	9s
10	0.0724	97.77%	0.0483	98.5%	99%	98%	99%	525.44	10508.8	0.0001	9s
11	0.0634	98.56%	0.0432	98.5%	99%	98%	99%	520.22	10404.4	0.0001	9s
12	0.0356	98.88%	0.0443	99%	99%	99%	99%	515.43	10308.5	0.0001	9s

Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.

Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.

Initial download of the metrics may take a while.