Lecture 2 What to do if my network fails to train

(一)局部最小值 (local minima) 与鞍点 (saddle point)
(二)批次 (batch) 与动量 (momentum)
(三)自动调整学习率 (Learning Rate)
(四)损失函数 (Loss) 也可能有影响
Lecture 2 What to do if my network fails to train_第1张图片Lecture 2 What to do if my network fails to train_第2张图片Lecture 2 What to do if my network fails to train_第3张图片Lecture 2 What to do if my network fails to train_第4张图片Lecture 2 What to do if my network fails to train_第5张图片Lecture 2 What to do if my network fails to train_第6张图片Lecture 2 What to do if my network fails to train_第7张图片Lecture 2 What to do if my network fails to train_第8张图片Lecture 2 What to do if my network fails to train_第9张图片Lecture 2 What to do if my network fails to train_第10张图片Lecture 2 What to do if my network fails to train_第11张图片Lecture 2 What to do if my network fails to train_第12张图片Lecture 2 What to do if my network fails to train_第13张图片Lecture 2 What to do if my network fails to train_第14张图片

你可能感兴趣的:(李宏毅,深度学习,神经网络,人工智能)