The platform provides predefined architecture presets that let you quickly configure a neural network without manually assembling layers. The following presets are available:
- Fully Connected Preset — A fully connected architecture composed of multiple Dense layers with intermediate Dropout layers. This preset is designed to improve generalization by reducing overfitting while maintaining strong learning capacity.
- Simple Fully Connected Preset — A compact architecture consisting of three stacked Dense layers with decreasing numbers of neurons. This preset is suitable for simpler tasks or scenarios where a lightweight model is preferred.