Create your own Tabular or Image Classification dataset, or load and modify/continue a previously created one.
If you already have a dataset in mind, skip this and create a neural network.
Create or Load/Modify your own neural network from scratch by selecting the type of network, number of layers,
number of neurons in each layer - and train it - then export it for deployment.
How It Works
If you don't already have a dataset but want to try it out, go ahead and CREATE a neural network, there is a sample
dataset you can play with (mnist handwritten digit recognition).
Otherwise, if you're starting from scratch:
1. Create a dataset (remember the type and number of inputs and outputs)
2. Create a neural network by matching the network configuration to the dataset configuration
(Image/Tabular, number of inputs/outputs, classification or regression)
3. Select the number of hidden layers and the number of neurons in each layer, along with their
activation functions (RELU by default).
4. Load a dataset, being carefull in regard to the configuration details.
5. Train the model, experimenting with different learning rates, as many epochs as you want.
6. Export the model (.json for model parameters and .bin for weights).
