Tagmentation of genomic DNA, PCR amplification with dual index primers, PCR clean-up using Agencourt AMPure XP (Beckman Coulter), DNA libraries normalization, library pooling and MiSeq sample loading were performed according to the manufacturer's instruction with minor modifications.
The multiple intermediate layers applied convolution, pooling, normalization, and other operations to capture the highly nonlinear mappings between inputs and outputs.
These iterated spatial operations – including pooling, normalization and pattern completion – enable these systems to recognize and predict spatial structure, while robust to changes in the spatial scale, contrast and noisiness of the input signal.
After purification of PCR fragments and library normalization, sample pooling was performed by mixing equal volumes of each normalized DNA library, and the barcoded multiplexed library sequencing was performed on an Illumina MiSeq (Illumina).
Triplicate reactions of each sample were combined after PCR and further processed as described in Kozich et al. [ 16], including between-sample normalization using the SequalPrep™ Normalization Plate Kit (Invitrogen GmbH, Darmstadt, Germany) and pooling of 96 samples.
We conclude that multiple consecutive stages of cortical processing can be understood to perform temporal pooling, temporal normalization and temporal pattern completion.
As network architecture, we use three convolutional layers with pooling, batch normalization, and non-linearities, followed by fully connected layers with dropout.
In addition, we perform our analysis on normalized array data and raw array data to examine how normalization affects pooling variance estimates.
