Event-Based Angular Velocity Regression with Spiking Networks (ICRA20 Video Pitch)

[@ahmadizzuddin]

Is there a decay on the potential over time if the threshold isn't reached?

[@amartinezsilberstein]

I haven't read the paper yet, but I understand that both in decreased and increased potential states there is a "decay" to the resting potential, as shown in the graph (1:55).

[@amartinezsilberstein]

Yes, the paper states that "whenever a spike arrives, the potential gets excited but decays again if no other spikes are registered close in time". This decay is exponential and depends on a set of hyperparameters.

[@jiachenghao1074]

of course it will, the potential will also be reset to a very low level if the threshold is reached.

[@arlenalem]

hello, It can works with moving object?

[@mathiasgehrig1286]

I have not investigated moving objects. The performance will likely be worse but you could augment the training dataset with moving objects to improve robustness

[@smithjack6626]

In the paper i notice that you used generated synthetic dataset . can you explain it?

[@zerotwo7319]

he explained at 3:27

[@ItsEndermanHD]

ag