0) The purpose of the recursive descent algorithm is to approximate the minimum value of the function by continually iterating, thus finding the parameter 1) the logistic regression is actually a classifier, using the existing sample to train the Sigmoid function.
(1) The general form of the sigmoid function:
(2) Graph of the sigmoid function:
(3) Prediction function:
For example, there is a sample X, he has 10 features:, according to the value of their predictive function can be obtained:
Then we can know the attribution of sample x: It is a class, otherwise it is another kind.
Note: This assumes a linear boundary condition: it is shaped like, not this. And the derivation is based on this hypothesis.
3) Derivation process
(1) First notice that the function domain is located in [0,1], the category is divided into 0, 12 classes.
The closer to 1, the more likely the sample belongs to Category 1; otherwise, the farther away from 1, the greater the likelihood that the sample will be in category 0.
So it can be seen as the features of a given sample X, and the known parameter θ, which belongs to the probability of Class 1:
, where the attribution category of the sample is represented by Y.
(2) Y satisfies two distributions
(3) According to MLE
(3) Define a new function:
(4) Derivation, to find the deviation of J (θ) to feature J
Define A, and B
So:
(5) Derivation of a, b
(6) Finally
Logistic regression & Recursive descent algorithm