The Input Layer shows how many input features are passed to the model for each inference. It is automatically estimated based on the following:
- Dataset columns.
- Enabled signal processing features.
- Window settings.
- Feature selection configuration.
When you change preprocessing or feature settings, this value updates automatically. The input layer does not require manual configuration. More input features may improve accuracy but also increase model size and memory usage, which is important for resource-constrained devices.
The Output Layer defines the final prediction format. This configuration is automatically determined by your task type:
- Regression: 1 neuron (linear activation).
- Binary classification: 1 neuron (softmax).
- Multi-class classification: N neurons (softmax), where N is the number of classes.
The output layer is set automatically and matches your selected task.