• ## Machine learning algorithm series (XIV) – hard margin support vector machine

Time：2022-3-30

Background knowledge points needed for reading this article: Lagrange multiplier method, KKT condition and yidui programming knowledge 1、 Introduction   in the previous section, we introduced a classification algorithm — naive Bayesian classifier algorithm, which classifies from the perspective of probability distribution. Next, we will spend a few sections to introduce another algorithm that plays […]

• ## Machine learning algorithm series (XV) – soft margin support vector machine

Time：2022-2-16

Background knowledge required for reading this article: hard interval support vector machine, relaxation variables, yidui programming knowledge 1、 Introduction    in the previous section, we introduced a most basic support vector machine model – hard interval support vector machine. This model can classify linearly separable data sets, but in reality, data sets are often linearly […]

• ## Machine learning algorithm series (XVI) – non linear support vector machine

Time：2022-2-14

Background knowledge required for reading this article: linear support vector machine and yidui programming knowledge 1、 Introduction    we introduced two support vector machine models in two sections earlier – hard interval support vector machine and soft interval support vector machine. These two models can be collectively referred to as linear support vector machine. Next, […]

• ## 04 Lagrange dual problem and KKT condition

Time：2021-12-24

04 Lagrange dual problem and KKT condition catalogue 1、 Lagrange dual function 2、 Lagrange duality problem 3、 Geometric interpretation of strong and weak duality 4、 Saddle point interpretation 4.1 basic definition of saddle point 4.2 minimax inequality and saddle point properties 5、 Optimality condition and KKT condition 5.1 KKT conditions 5.2 KKT condition and convex […]

Time：2021-12-16

08-admm algorithm catalogue 1、 ADMM algorithm motivation 2、 Dual problem 3、 Dual ascent method 4、 Dual partition 5、 Multiplier method (augmented Lagrange function) 5.1 benefits of \$\ Rho \$in steps 6、 ADMM algorithm 6.1 scaled form of ADMM 7、 Proof of convergence of ADMM 8、 Write at the end Convex optimization from getting started to […]

• ## Exercises in Chapter 7 of statistical learning methods

Time：2021-12-8

Exercise 7.1 The difference between the dual form of perceptron and that of support vector machine is that perceptron is transformed by assuming the increment of parameter change; The support vector machine is solved by solving the constrained optimization problem, which is transformed into an unconstrained optimization problem through Lagrange duality. The original form of […]

• ## Machine learning: principle derivation of SVM

Time：2020-10-11

It is said that SVM is the watershed of machine learning, machine learning is just around the corner. This paper will introduce the principle derivation process of SVM in detail, including linear, near linear, nonlinear, optimization methods, etc. a lot of ideas are derived from statistical learning method and zero basis introduction Python data mining […]

• ## White board derivation of SVM | derivation process of loss function evolved from maximum interval target

Time：2020-9-28

The English name of SVM is support vector machine, which is called support vector machine. The perceptron learning algorithm will get different hyperplanes because of different initial values. However, SVM tries to find an optimal hyperplane to divide the data. How to calculate the best one? We can naturally think that if the distance from […]

• ## KKT and KKT support vector machine learning

Time：2020-7-29

Overview of SVM Support vector machine (SVM) is a supervised classification algorithm, and it mostly deals with the problem of binary classification. First of all, through a series of pictures to understand several concepts about SVM. In the above figure, there are orange dots and blue dots representing the two types of labels. If you […]