Networks of Spiking Neurons Learn to Learn and Remember

Wolfgang Maass, Graz University of Technology
simons.berkele...
Computational Theories of the Brain

[@joaomonteiro2286]

Good video. A list of all acronyms would be really helpful though.

[@ChristopherHarris]

Great video. Is the subsequent discussion available to watch?

[@maxreeder]

I believe this is the paper, if anyone is interested: arxiv.org/pdf/1803.09574.pdf

[@bikinihaulsell_bikinis3125]

No use to make snn do the same thing as LSTM like this

[@thomasmuller7001]

why?

[@bikinihaulsell_bikinis3125]

@@thomasmuller7001 LSTM are very good at sequence learning but this is not the essence of what spiking networks do, imho they try to internalize behavioural decision factors. They are also not sequentially locked like LSTMs but depend on timing of stimulus and internal state..LSTM is a special case of what spiking networks in de brain do imho

[@emademad4]

@@bikinihaulsell_bikinis3125 yes they are good , but for big real world complicated task you would need a very very big net of them and training would be impossible and hence not useful for real time operation