Fig. 9: Theoretical predictions.
From: Separability and geometry of object manifolds in deep neural networks
a Comparison of numerically measured capacity (x-axis) with the theoretical prediction (y-axis) for AlexNet, VGG-16 at different layers along the hierarchy (top 10% point-cloud manifolds). b Comparison of numerically measured capacity (x-axis) with the theoretical prediction (y-axis) for AlexNet at different layers along the hierarchy and different levels of manifold variability (smooth 2-d manifolds). c Numerically measured capacity (y-axis) at different number of objects (x-axis) for point-cloud manifolds at different layers (dashed line: top 10% manifolds; dotted line: top 5% manifolds). d Numerically measured capacity (y-axis) at different number of objects (x-axis) for smooth 2-d shear manifolds. Marker shape represents layer type (circle—pixel layer, square—convolution layer, right-triangle—max-pooling layer, hexagon—fully connected layer, down-triangle—local normalization layer). Color (blue—AlexNet, green—VGG-16) changes from dark to light along the network.
