Big Transfer Learning for Fine Art Classification

Algorithm 1

Fine-tuning the neural network and obtaining the painting classification accuracy.

	Input: Initialize(net)
	Output: Test Acc
(1)	while 1 do
(2)	for batch = 1,2, …,#images/do
(3)	images uniformly random sample b images
(	progress(images);
(5)	get lr(, step);
(6)	mixup data
(	forward
(	loss
(	grad backward
(	update(net,grad);
(11)	step++;
(12)	ifthen
(13)	break;
(14)	end
(15)	end
(16)	end
(17)	Acc = Eval(net, test)