30 suggests a potentially interesting relationship between DeepClean's effective receptive field and the bandwidth of signals it can model. If this is true, it would be cool to try to derive what this tradeoff ought to be analytically from the uncertainty principle and then validate it on either synthetic or real data (or both).
30 suggests a potentially interesting relationship between DeepClean's effective receptive field and the bandwidth of signals it can model. If this is true, it would be cool to try to derive what this tradeoff ought to be analytically from the uncertainty principle and then validate it on either synthetic or real data (or both).