Teaching a neural network oriented computational modeling class this semester, and I'm trying to implement Elman (1990) in Keras. Getting the "simple" XOR network to work in experiment 1 is hard!
This pattern is also puzzling to me. How can the error on the unpredictable steps be anything but 0.5?
September 19, 2023 at 3:32 PM
Teaching a neural network oriented computational modeling class this semester, and I'm trying to implement Elman (1990) in Keras. Getting the "simple" XOR network to work in experiment 1 is hard!
This pattern is also puzzling to me. How can the error on the unpredictable steps be anything but 0.5?