Methodology.Convergence, 3p, This suggests that a sufficiently large number of total samples (arguably in terms of pixels) are required for the models trained from random initialization to converge well. (실험을 통해, 모델에 스쳐지나간 픽셀의 수가 근접해지면, pretrained weight + fine-tuning 된 model 과 from scratch 된 model 의 성능이 비슷해진다고 결론내리고 있다.) 실험*