Medical Text Classification Using Hybrid Deep Learning Models with Multihead Attention

Table 13

Analysis of results with different activation functions for the proposed hybrid BiGRU model.

Optimizer	Dataset	Accuracy (%)	Precision (%)	Recall (%)	F1-score (%)
ReLU	Hallmarks	71.64	67.39	66.69	67.03
	AIM	88.39	82.20	83.61	82.89
Sigmoid	Hallmarks	74.69	72.29	71.91	72.09
	AIM	94.29	89.74	88.56	89.14
SoftPlus	Hallmarks	73.12	71.03	71.82	71.42
	AIM	83.39	81.91	82.84	82.37
Hard sigmoid	Hallmarks	61.22	58.92	57.41	58.15
	AIM	85.87	83.19	81.49	82.33