Table 3 Overall CTR prediction for RMSE performance in different datasets.
From: Session interest model for CTR prediction based on self-attention mechanism
Model | RMSE | |||
|---|---|---|---|---|
Books | Electronics | Avazu | Criteo | |
Wide&Deep | 0.5071 | 0.5224 | 0.5326 | 0.6072 |
FNN | 0.5043 | 0.5202 | 0.5302 | 0.5986 |
AFM | 0.4996 | 0.5138 | 0.5279 | 0.5823 |
DeepFM | 0.4942 | 0.5079 | 0.5213 | 0.5761 |
DIN | 0.4578 | 0.4693 | 0.5185 | 0.5625 |
ADI | 0.3675 | 0.4267 | 0.5061 | 0.5485 |
SISA | 0.3121 | 0.3783 | 0.4925 | 0.5306 |
