Table 4 Classification accuracy besides running period for image block input representations.
From: Hybrid optimization driven fake news detection using reinforced transformer models
Dataset | Model Name | Input Format | OA (%) | Kappa | Train Time | Test Time |
|---|---|---|---|---|---|---|
DGM4 | CNN | Image block | 79.81 | 0.7215 | 100 m33 s | 12 s |
Proposed | 86.11 | 0.8063 | 108 m 13 s | 14 s | ||
CNN + LSTM | 85.86 | 0.805 | 128 m 38 s | 12 s | ||
Fakeddit | CNN | Image block | 85.93 | 0.8034 | 132 m | 13 s |
CNN + LSTM | 90.76 | 0.8562 | 131 m 44 s | 15 s | ||
Proposed | 91.21 | 0.8615 | 239 m 46 s | 14 s |
