fig4

Figure 4. Deep learning-based framework (CNN + RNN) for kirigami design^[39]. (A) Kirigami dataset and neural network modeling, where CNNs classify geometric configurations and buckling modes to extract key design parameters; (B) Parameter design through forward and target-driven approaches to optimize geometric configurations for enhanced mechanical behaviors such as improved buckling modes and stability; (C) Evolving kirigami structures for specific functionalities, enabling tailored designs for improved tensile performance, controlled buckling behavior, and expanded application potential. CNN: Convolutional neural network; RNN: recurrent neural network.