No abstract

No abstract

No abstract

1AbstractWe hypothesize that deep networks are superior to linear decoders at recovering visual stimuli from neural activity. Using high-resolution, multielectrode Neuropixels recordings, we verify this is the case for a simple feed-forward deep neural network having just 7 layers. These results suggest that these feed-forward neural networks and perhaps more complex deep architectures will give superior performance in a visual brain-machine interface.

A new parameterized window function over a fixed support is presented in this work. It is derived through the process of iterative sidelobe inversion, which is used to achieve a window with optimal sidelobe properties for any given main lobe width. Sidelobe inversion is presented as an operator on window functions, which results in new window functions of the same length over the same finite fixed support. For any given set of windows with the same main lobe width in the frequency domain, it is shown that iterative application of sidelobe inversion converges to a fixed window function. This new window can thus be created for any main lobe width and has small peak sidelobes and reduced total side lobe energy, outperforming many of the known window functions.