RCSB PDB Protein Structure Comparison - TensorBoard Logs

This archive contains TensorBoard log files generated during training and evaluation of a fully connected neural network for the SDSC RCSB PDB protein comparison task. The model was trained using BioZernike descriptors (geometric and Zernike moments) derived from 3D protein structures to classify structural similarity between protein pairs. Logs include metrics such as: ROC AUC (Receiver Operating Characteristic) PR AUC (Precision-Recall Area Under Curve) MCC (Matthews Correlation Coefficient) Loss curves and weight histograms The training was conducted on the cath_moments.tsv dataset, and performance evaluation was done using ecod_moments.tsv. Logs are formatted for TensorBoard and can be visualized using the TensorBoard dashboard for further analysis of the training process across different model sizes (1, 32, 64, 128, and 256,512 and 1024 neurons)