Predictions on live data

Hello. I’m trying to predict where a particle is going to move. I have the current location of the particle, and I’d like to feed that to the tensor, have it predict the next location, and then compare this prediction to the particles actual location and update based on if the machine made the correct prediction. I don’t want to store the data if I can avoid it. Is tensor flow capable of this?

You will typically need to train the TensorFlow model using many data points before making predictions. But I suppose it’s possible to create a “null” model (or “train” a model with one fake data point), make a prediction, train it with that (real) data point, make another prediction, and so on.

Maybe you can simultaneously predict the position and momentum of a quantum particle. :slight_smile: