Table 2 Classification of neural networks by network structure.

From: Deep learning driven prediction and comparative study of surrounding rock deformation in high speed railway tunnels

Typology

Applicable Scenarios

Specificities

Subcategories

FNN

Simple classification and regression tasks.

Information flows unidirectionally from the input layer to the output layer without feedback connections

MLP, Autoencoder

CNN

Image classification, object detection, segmentation, etc.

Convolutional operations extract local features, making them suitable for image data processing

VGG, TCN

RNN

Natural language processing and time series forecasting.

Utilizes time-memory features for processing time-series data

LSTM, GRU

GAN

Image generation and style transfer.

Composed of a generator and a discriminator, which generate data through adversarial training

cGAN, WGAN

GNN

Graph classification, node prediction, and edge prediction.

Designed for processing graph-structured data, such as social networks and molecular structures

GCN, GAT

Attention Mechanism

Natural language processing and image-related tasks.

Selects more important input information using attentional mechanisms

Transformer, Self-Attention Network

Back to article page