Sentence examples for decoder patterns from inspiring English sources

As for the hemispheric decoder, patterns were computed in the same frequency band in which the decoder is tested.

In the context of the HNN decoder, the patterns are the balanced codewords, and the HNN is able to determine the ML codeword from a set of codewords.

The difference between hemispheric decoder and pattern decoder only appears in this more general situation.

We also consider a frequency-dependent pattern matching decoder in which the pattern ρ j is broken into subvectors corresponding to frequency bands, and each subvector is normalized separately.

The first method is to resynchronize the video decoder using resynchronization patterns.

Some of the decoders used in these studies are, indeed, similar to the pattern decoder we used (the closest one being the maximum likelihood decoder in Miller and Recanzone, 2009), however we do not agree that these studies do empirically what we are doing with simulations.

We also added a mention of the error of the pattern decoder in the first paragraph.

One may immediately object that the pattern decoder is in fact too accurate compared to natural performance, in particular for humans.

The performance of the pattern decoder is generally not strongly dependent on frequency in the 500 1200 Hz range, whereas the performance of the hemispheric decoder consistently increases with frequency (between about 500 and 1 kHz in Figure 2).

The pattern decoder is essentially a perceptron (Dayan and Abbott, 2001): spatially tuned neurons are formed by simply pooling neural responses with different weights, and then the maximally active neuron indicates source location.

For example, changes in various aspects of the sound induce shared variability in individual responses, which results in the same variability in pooled responses, but little variability in the relative activity of neurons (used by the pattern decoder).