机器学习-算法理论

模型的variance&bias

2020-07-28  本文已影响0人  shudaxu

Bias:Bias are the simplifying assumptions made by a model to make the target function easier to learn.
Bias refers to assumptions in the learning algorithm that narrow the scope of what can be learned. This is useful as it can accelerate learning and lead to stable results, at the cost of the assumption differing from reality.

通俗点来讲,Bias就是模型本身的假设限制,带来的偏差。

Variance:Variance is the amount that the estimate of the target function will change if different training data was used.
Variance refers to the sensitivity of the learning algorithm to the specifics of the training data, e.g. the noise and specific observations. This is good as the model will be specialized to the data at the cost of learning random noise and varying each time it is trained on different data.
Variance就是当我们更换不同的训练集时,模型(target function中参数)的变化。*直觉上理解就是,模型能够挖掘选取出数据背后的“真理”,那么在数据变化后,其“真理”仍然在,所以切换数据后,不会有很大的差异。variance小则模型Less sensitive(模型稳定性强)

fx为真实值,f‘x为一次的预估值,左侧的点为多次训练实验后的平均结果。可以看到,bias error和variance error共同构成了泛化误差(generalization error),当然,其实数据和真实之间还有Noise,但是不在当前的讨论中,因为Noise只能由数据的处理去解决,数据决定了上限。PS:当然,训练数据中的Noise也会造成Variance的增大(真理不是那么显著了)

我们的目标当然是缩小泛化误差,路径就是缩小bias的同时缩小variance。但是由上述分析可知,减小variance则会增大bias(简单模型,复杂模型)(反之亦反),所以这里才有了bias and variance的trade-off。当然,在模型训练中,我们有很多手段可以调节这个trade-off的度

如何降低模型的variance:https://machinelearningmastery.com/how-to-reduce-model-variance/

注:

上一篇下一篇

猜你喜欢

热点阅读