-
Enhance the system by integrating widget components to visualize training data, enabling the display of network topology, bar charts, and forecasts. This feature expands beyond the limitations of an active wallpaper, offering a more interactive and versatile way to plot and interact with training information.
-
Add an option to activate the application as a widget or as wallpaper.
-
Separate wallpaper service as calculating service and visualization service.
-
Add the possibility for the user to mark his guess for forecasting and collect this information on the server. Some users can be very successful in the guessing of the price movements.
-
Use the sine function with a very wide period as a neuron's activation function instead of the sigmoid function. The cosine is the derivative of the sine, and backpropagation can be used for the training.
-
On a regular basis, multiply all weights by 1/100. In this way, ANN will cool down and proceed with training.