Fig. 1. Schematic diagram of micro-resonator system

Fig. 2. Air hole array etched waveguide micro-resonator system

Fig. 3. Transmission spectrum of unetched micro-resonator

Fig. 4. MLP Network training process

Fig. 5. Loss comparison across neural networks with varying layer counts

Fig. 6. Training and validation losses at LR 0.000 1

Fig. 7. MRR transmittance simulation using neural network and FDTD method

Fig. 8. Prediction curves of transmission spectra for air hole array and corresponding micro-resonator

Fig. 9. Comparison of model cosine similarity at different epochs

Fig. 10. Model error evaluation chart

Fig. 11. Data encoding and decoding transformation

Fig. 12. Inverse design and verification process

Fig. 13. Grayscale plot of predicted transmission characteristics for different micro-resonator

Fig. 14. Changes in slope and design parameter differences during the optimization process

Fig. 15. Relationship between slope and centroid position during optimization process