Sentence examples for the classification image from inspiring English sources

Tuning of the classification image profiles changes remarkably little with stimulus size.

To ensure that the classification image profiles are not distorted by this difference of spatial areas, we computed normalized classification images by dividing each point in the classification image by the area of the stimulus ring (number of the pixels).

The classification image reveals how much total weight the subjects give to the stimulus information in different rings.

We also point out that a spreading of the classification image profile does not necessarily imply low-frequency boosting.

Our results are very different; the classification image profile of the illusory stimulus was not wider than target.

On the other hand, this normalization makes the classification image profiles very noisy in the most central points (rings) that had area of only a few pixels.

Therefore we analyzed also the weights per unit area by normalizing the classification images by the area of each ring.

The standard deviation of the Bootstrap replicas was used as an estimate of the standard error of the classification images.

In fact, the classification images are very similar for both real (step edge) and illusory (Craik-O'Brien-Cornsweet) stimuli.

In summary, we found that (1) profiles of the classification images for perceived brightness peak at the border of the patch.

To characterize further the tuning, odd-symmetric exponential functions were fitted to the excitatory lobes of the normalized classification image profiles.