Is there any mathematical equation to determine the appropriate number of hidden layers for a sequential model?
Hi Hasintha, It's a practical problem and you should make decision about number of hidden layers by checking accuracy of the fitted model.
I agree with Seyed Mahdi Amir Jahanshahi , empirical evaluation of alternative specifications is the key to a reliable solution here.
Please let me recommend this work describing a framework allowing the implementation of the empirical analysis:
Chapter Forecast Evaluation Techniques for I4.0 Systems
The chapter is largely based on this review:
Chapter Measuring Forecasting Accuracy: Problems and Recommendations...
Usually Number of hidden layers , optimizers, activation functions etc. are all considered to be the hyper-parameters of a specific neural network model. For a given model architectuere(sequential or non-sequential), one can find the optimal values of HPs through some HP tuning mechanism against some accuracy measure(s).
- Badges
- Science topic
- Similar topics
- Mathematics