Skip to main content
. 2022 May 6;16:883693. doi: 10.3389/fnins.2022.883693

TABLE 1.

The network architecture of the MST-net model.

MST (5) MST (7) MST (9)
Input (2 × 1,000)
Stream 1 Stream 2 Stream 3
Layer 1 Conv (15)
Layer 2 Max-pooling (3)
Layer 3 Conv (5)-64 Conv (7)-64 Conv (9)-64
** ** **
Layer 4 Conv (5)-64 Conv (7)-64 Conv (9)-64
** ** **
Layer 5 Conv (5)-128 Conv (7)-128 Conv (9)-128
** ** **
Layer 6 Conv (5)-128 Conv (7)-128 Conv (9)-128
** ** **
Layer 7 Conv (5)-256 Conv (7)-256 Conv (9)-256
** ** **
Layer 8 Conv (5)-256 Conv (7)-256 Conv (9)-256
** ** **
Layer 9 Conv (5)-512 Conv (7)-512 Conv (9)-512
** ** **
Layer 10 Conv (5)-512 Conv (7)-512 Conv (9)-512
** ** **
Layer 11 AvgPool1d (1) AvgPool1d (1) AvgPool1d (1)
Layer 12 FC-512 FC-512 FC-512
Layer 13 FC-256
Output-3

**Represents “Batch Normalization layer + non-linear function rectifier linear unit.”