摘要:
Reference https://developers.google.com/machine-learning/crash-course/generalization/peril-of-overfitting https://developers.google.com/machine-learni 阅读全文
摘要:
原创转载请注明出处:https://www.cnblogs.com/agilestyle/p/12620570.html 偏差 Bias 反映的是模型在样本上的输出与真实值之间的误差,即模型本身的精准度。 方差 Variance 反映的是模型每一次随数据集的变动造成的输出结果与模型输出期望之间的误差 阅读全文
摘要:
Reference: https://developers.google.com/machine-learning/crash-course/reducing-loss/learning-rate Learning Rate 正如之前所述,梯度矢量具有方向和大小。梯度下降法算法用梯度乘以一个称为学习 阅读全文
摘要:
Reference: https://developers.google.com/machine-learning/crash-course/reducing-loss/gradient-descent Gradient Descent 迭代方法图 包含一个标题为“计算参数更新”的华而不实的绿框。现 阅读全文
摘要:
Reference: https://developers.google.com/machine-learning/crash-course/reducing-loss/an-iterative-approach iterative approach 机器学习算法用于训练模型的迭代试错过程: 我们将 阅读全文
摘要:
Reference: https://developers.google.com/machine-learning/crash-course/descending-into-ml/training-and-loss 简单来说,训练模型表示通过有标签样本来学习(确定)所有权重和偏差的理想值。在监督式学 阅读全文
摘要:
Reference: https://developers.google.com/machine-learning/crash-course/descending-into-ml/linear-regression linear regression 每分钟的虫鸣叫声与温度(摄氏度)的关系 此曲线图 阅读全文