Simulate collaborative ML scenarios, experiment multi-partner learning approaches and measure respective contributions of different datasets to model performance.
Compare to what we can found in the literature, our cifar model converge very slowly. We should be able to reach 70%, 80% accuracy, in relatively short amount of epoch.
using adams, I reached 70% in 40 epochs.
with 10 times higher learning rate with rmsprop, 60% accuracy in 40 epochs
Compare to what we can found in the literature, our cifar model converge very slowly. We should be able to reach 70%, 80% accuracy, in relatively short amount of epoch.
using adams, I reached 70% in 40 epochs. with 10 times higher learning rate with rmsprop, 60% accuracy in 40 epochs