Skip to main content

Table 2 Sch-net architecture details

From: Sch-net: a deep learning architecture for automatic detection of schizophrenia

Layer Dimension
Conv1 2×[3×3(64 filters)]
Conv2 2×[3×3(128 filters)]
Conv3 2×[3×3(256 filters)]
Conv4 2×[3×3(512 filters)]
Conv5-8 3×3(512 filters)
Max-pooling 2×2
Average-pooling 2×2
CBAM 7×7 (2048 filters)
FC 1×1×512, 1×1×2 (two hidden layers)