|
it
Pav = 1/N Σ Pi, Disp = 1/N Σ(Pi - Pav)2, Error = (Disp/N)1/2. |
"6 Daisies bouquet" with cubic symmetry is used. Press "iterate" to collect 1000 daisy-class (985) probabilities. See console for more detailes. CNN "GraphModels" from tfhub.dev are used.
CNN Pav Daisy Pav Bouquet T ms ________________________________________ v2_0.35 0.29 0.28 18.8 v2_0.50 0.54 0.37 19.5 v1_1.00 0.74 0.48 24.5 v1_1.00* 0.48 v2_1.00 0.47 0.56 24.7 v2_1.40 0.56 0.41 30.7 ResNet 0.74 0.95 59.2* I get Pav = 0.48 for the same v1_1_224 "LayersModel" at 'https://storage.googleapis.com/tfjs-models/tfjs/mobilenet_v1_1.0_224/model.json'. See "Daisy" weight quantization test.
Times are measured on HP Laptop with AMD Ryzen 5 3500U + Vega8.
"Daisy" Bouquet - classification test is much better than the single "Daisy" one. If some flowers fall into the blind zone, then we see contribution from the rest!