Fig. 1. Perceptron artificial neural network for phase retrieval^[18]

Fig. 2. Modified Inception v3CNN model for predicting Zernike coefficients^[20]

Fig. 3. Zernike coefficients predicting results of focused target^[23]

Fig. 4. Zernike coefficients predicting results of overexposed target^[23]

Fig. 5. WFSless system architecture^[26]

Fig. 6. Architecture of CNN^[26]

Fig. 7. Data flow of training and predictions^[29]

Fig. 8. Residual wavefront RMS with and without compensation under different turbulence levels^[29]

Fig. 9. Zernike coefficients prediction results by models trained with dataset of different turbulence levels^[31]

Fig. 10. Trained neural network is optimized by TensorRT to build the inference engine for implementation^[35]

Fig. 11. Wavefront Net (WFNet)^[36]

Fig. 12. CNN architecture^[38]

Fig. 13. Standard deviation of phase before and after phase aberration revision^[38]

Fig. 14. An object irrelevant wavefront sensing scheme using LSTM neural network^[41]

Fig. 15. Image restoration results based on wavefront error inferred by LSTM^[41]

Fig. 16. Prediction results of the next 5 frames wavefront made by LSTM^[44]

Fig. 17. Reinforcement Learning(RL) of WFSless AO^[47]

Fig. 18. Intensity distribution of point target with wavefront error and that after restoration by deep RL^[47]

Fig. 19. PSD of different vibration frequency^[51]

Fig. 20. Residual phase^[51]

Fig. 21. Principle of aberration correction in high resolution optical microscopes^[59]

Fig. 22. Schematic diagram of MAL-WSAO-SS-OCT system^[63]