"How many hidden layers should be used in Neural Net?"
Is there any method by which I can check that this many Hidden layers and Neurons are best suitable for this model?
Find more posts tagged with
Sort by:
1 - 1 of
11
Sort by:
1 - 1 of
11
Hi @vjmeena
There is no thumb rule for a number of layers in neural networks. In my works, I generally build complex neural networks (with more layers) if there are many data samples. For, small datasets neural net with fewer layers would suffice. Try different combinations and observe the performance metrics. Based on that you can see which is best.
You can also try AutoMLP which gets an optimal number of neurons based on training different networks in parallel.
Performance of neural nets is mainly influenced by parameters like activation functions, learning rates, drop out etc.
There is no thumb rule for a number of layers in neural networks. In my works, I generally build complex neural networks (with more layers) if there are many data samples. For, small datasets neural net with fewer layers would suffice. Try different combinations and observe the performance metrics. Based on that you can see which is best.
You can also try AutoMLP which gets an optimal number of neurons based on training different networks in parallel.
Performance of neural nets is mainly influenced by parameters like activation functions, learning rates, drop out etc.
There is no thumb rule for a number of layers in neural networks. In my works, I generally build complex neural networks (with more layers) if there are many data samples. For, small datasets neural net with fewer layers would suffice. Try different combinations and observe the performance metrics. Based on that you can see which is best.
You can also try AutoMLP which gets an optimal number of neurons based on training different networks in parallel.
Performance of neural nets is mainly influenced by parameters like activation functions, learning rates, drop out etc.