Deep Learning Specialization

Notes from DeepLearning.AI Specialization in Deep Learning by Andrew Ng

Neural Networks and Deep Learning

Logistic Regression

Log Loss vs Mean-Squared Error

M.S.E(y,y′)=∑i=1n(yi−yi′)2M.S.E(y,y') = \sum _{i = 1} ^{n} (y _i - y' _i)^2
LogLoss(y,y′)=−∑i=1nyi∗log(yi′)LogLoss(y,y') = -\sum_{i=1}^{n} y _i * log(y' _i)

Log Loss is a convex function whereas Mean-Squared-Error is a non-convex function with multiple local optima and thus might never converge to a global minima.

Neural Networks

Improving Deep Neural Networks

Structuring Machine Learning Projects

Convolutional Neural Networks

Sequence Models

PreviousIntroductionNextTensorflow

Last updated

Was this helpful?