Figure 8
The vertical axis is labeled “Percentage Improvement in Accuracy (percent)” and ranges from negative 2.00 percent to 18.00 percent in increments of 2.00 percent. The horizontal axis is labeled “D I Models” and lists categories: “C N N-B R S”, “Bays-C N N”, “C N N-W R S”, and “Cybernet”. At the top the legend includes four entries: “C I C-D D o S 2019 C N N-L S T M” shown in blue bars, “C I C-D D o S 2019 C N N” shown in orange bars, “S D N-D D o S 2025 C N N-L S T M” shown in gray bars, and “S D N-D D o S 2025 C N N” shown in yellow bars. The numerical values are written at the top of each bar. The data for the bars are as follows: C N N-B R S: C I C-D D o S 2019 C N N-L S T M: negative 0.01 percent; C I C-D D o S 2019 C N N: negative 0.01 percent; S D N-D D o S 2025 C N N-L S T M: 6.69 percent; S D N-D D o S 2025 C N N: 4.24 percent. Bays to C N N: C I C-D D o S 2019 C N N-L S T M: 0.08 percent; C I C-D D o S 2019 C N N: 0.08 percent; S D N-D D o S 2025 C N N-L S T M: 11.04 percent; S D N-D D o S 2025 C N N: 7.85 percent. C N N-W R S: C I C-D D o S 2019 C N N-L S T M: 4.22 percent; C I C-D D o S 2019 C N N: 4.22 percent; S D N-D D o S 2025 C N N-L S T M: 16.69 percent; S D N-D D o S 2025 C N N: 12.11 percent. Cybernet: C I C-D D o S 2019 C N N-L S T M: negative 0.01 percent; C I C-D D o S 2019 C N N: negative 0.01 percent; S D N-D D o S 2025 C N N-L S T M: 5.24 percent; S D N-D D o S 2025 C N N: 2.97 percent.Percentage improvements in classification accuracy of the proposed CNN, CNN-LSTM models relative to benchmark architectures across the CIC-DDoS2019 and SDN-DDoS2025 datasets