is that possible to create an spiking auto-encoder with nengo about nengo-dl HOT 4 CLOSED

There's an example of implementing an autoencoder in Nengo here https://www.nengo.ai/nengo-dl/examples/from-tensorflow.html

from nengo-dl.

Yes, you would need to change the neuron model from nengo.RectifiedLinear to a spiking neuron model (e.g. nengo.SpikingRectifiedLinear), and then you would need to make sure that you are running the network for multiple timesteps during inference/evaluation. Basically the same as this example https://www.nengo.ai/nengo-dl/examples/spiking-mnist.html, but you're training an autoencoder instead of a classification network. But there's nothing qualitatively different about training an autoencoder, it's just a different network structure and training data.

from nengo-dl.

Thank you very much for your response @drasmuss
but one question
this link that you sent me ,is an autoencoder in spiking neural networks??
i guess it's not
and if it's not , do you have any idea how to convert it to an autoencoder in SNNs??

from nengo-dl.

Closing this since it seems like the question is answered, but if you have any other questions feel free to stop by the forum and we can help out!

from nengo-dl.