Fig. 4: Transition rates estimation of COCOC and CCCOO models using neural networks.
From: A deep learning approach to real-time Markov modeling of ion channel gating
The regression architecture (Fig. 1A) was trained using datasets No. 2 and 3 (Table 1) containing models of the COCOC and CCCOO topologies, respectively. A–D After training, the network was evaluated using the test dataset. A, B illustrate the results of the predictions for the overall best-predicted rates k54, k34, and C, D for the worst-predicted rates k21, k21 according to the overall RAE score for topologies COCOC and CCCOO, respectively. Each test dataset contains 10,000 samples (2D-histograms). The orange short-dashed line and the red dashed line indicate the points on the graphs which have error scores (RAE) equal to 0.6 and 1.0, respectively.
