TABLE 1.
Hyperparameter settings used for training the network with and without receptive fields.
| Hyperparameter | Meaning | Value (with RFs) | Value (without RFs) |
| N | Number of layers | 4 | 4 |
| sl, ∀l ∈ {1, 2, 3, 4} | Size of receptive fields | 7 | Fully connected |
| n 1 | Population size (Number of neurons in a population) in area 1 | 8 | 5408 |
| n 2 | Population size in area 2 | 16 | 6400 |
| n 3 | Population size in area 3 | 32 | 6272 |
| n 4 | Population size in area 4 | 64 | 4096 |
| γy | Update rate for inference | 0.05 | 0.0005 |
| γw | Learning rate for synapses | 0.05 | 0.0005 |
| αy | Regularization for causes | 0.001 (all areas) | 0.0001 |
| αw | Regularization for weights | 0.001 (all areas) | 0.001 |
The size of receptive field in the network with receptive fields is equal in both image dimensions. Note that the term receptive field (RF) has been used in this table in line with its conventional definition. For the network without RFs, n1, n2, n3, and n4 are equal to the total number of neurons in each area.