fig4
From: Integrating sequence and chemical insights: a co-modeling AI prediction framework for peptides
Figure 4. Comparison of classification accuracy between two non-co-modeling baselines (T: Transformer; G: GraphSAGE) and co-modeling implementations (WS: weighted sum; Concat: concatenation; CA: cross-attention; CBP: compact bilinear pooling; RepCon: representation contrasting) on the datasets AMP and PepDB. Each box represents the distribution of accuracy for a method, including the minimum, first quartile (bottom of the box), median (line within the box), third quartile (top of the box), and maximum. GraphSAGE: Graph sampling and aggregation networks; AMP: anti-microbial peptide; PepDB: Peptide DataBase.
