Convolutional neural network model for healthcare prediction tasks.
Each feature is embedded and processed through independent CNN layers.
The pooled representations are concatenated for final prediction.
Usage
CNN(dataset = NULL, embedding_dim = 128, hidden_dim = 128, num_layers = 1)
Arguments
- dataset
A SampleDataset object providing input/output schema
- embedding_dim
Integer, embedding dimension. Default 128
- hidden_dim
Integer, number of CNN channels. Default 128
- num_layers
Integer, number of CNN blocks. Default 1
Details
Supports binary, multi-class, and regression tasks
Each feature has its own CNN encoder
Handles sequence, timeseries, and tensor inputs
Automatically determines spatial dimensions based on processor type