Page 24 - Read Online

Liu et al. J Mater Inf 2024;4:33 https://dx.doi.org/10.20517/jmi.2024.48 Page 17 of 33

Table 6. Typical deep learning models with their features and applications
Deep learning model Features Applications
DNN Multi-layer architecture Classification, regression
CNN Convolutional layers for spatial data Image classification, object detection
RNN Sequence modeling Time-series, speech recognition
LSTM Long-term dependency handling Time-series forecasting
GAN Adversarial training Image generation, data augmentation
GNN Graph structure processing Social networks, drug discovery
Transformer Self-attention mechanism NLP, machine translation
SHAP Feature importance attribution Model interpretability, finance
Grad-CAM Visual attention mapping Model transparency, medical imaging
KAN Universal function approximation Nonlinear system modeling

DNN: Deep Neural Network; CNN: Convolutional Neural Network; RNN: Recurrent Neural Network; LSTM: Long Short-Term Memory; GAN:
Generative Adversarial Networks; GNN: Graph Neural Networks; SHAP: shapley Additive explanations; Grad-CAM: Gradient-weighted Class
Activation Mapping; KAN: Kolmogorov-Arnold Network.

Figure 7. Machine learning aided discovery of novel photocatalysts. (A) Machine learning aided design of perovskite oxide materials for
[126]
photocatalytic water splitting . Copyright 2021, Elsevier; (B) Machine learning strategy for designing high-performance photoanode
[127]
catalysts . Copyright 2023, Royal Society of Chemistry.