MFDA: Multiview fusion based on dual-level attention for drug interaction prediction.

After that, all of them had a decreasing trend, indicating that too many layers may also incorporate noisy information, which instead reduced the model effect.

also in 47,180 other papers

The results are shown in Figure 5C , where the model performance showed an increasing trend until it reached three levels, indicating that it incorporated more useful information.

also in 63,971 other papers