search

chan8972 / Enabling_Spikebased_Backpropagation

MIT License
32 stars 7 forks source link

Issue about model training on N-MNIST #4

Open TJXTT opened 3 years ago

TJXTT commented 3 years ago

Hi,

I'm trying to implement LeNet using the N-MNIST dataset but the accuracy does not increase as the training step growing. I load the N-MINST dataset by PySNN ( https://github.com/BasBuller/PySNN).

Can you please tell me how to train the model on N-MNIST?

Your help will be appreciated. Thanks

chan8972 commented 3 years ago

We are not aware of PySNN code. We have preprocessed the N-MNIST dataset (similar to "https://github.com/chan8972/Spike-FlowNet/blob/master/encoding/split_coding.py") and use dataloader to feed the data through the network. Since N-MNIST contains sparse event data, there typically requires a higher degree of variance for the weight parameter initialization.