Neural Network Square.png

An animation asking “How does a neural network work?”. The animation shows precipitation and air temperature as inputs to a neural network predicting stream temperature. The input data is passed through weighted connections to nodes in a hidden layer that calculate weighted sums. The final output layer produces a prediction of stream temperature. The difference between the prediction and the actual temperature is minimized by repeatedly changing the weights in the model until the model produces the most accurate predictions based on the data.