Machine learning Personal Insights----late-night summary

Today, again, Wunda's machine learning program, suddenly felt like suddenly understand what is machine learning

In fact, we are all "learning" the two words scare, misleading, subconsciously, the machine as a person to see, naïve think, the machine will want to let people think, will learn, this is the biggest misunderstanding, become our study on the road of the biggest misleading, but very few people will know.

In fact, machine learning is more like an algorithm, and this algorithm is how to get it, through the program you write to continue to input data experience, the algorithm must be through the external data to continue to implement, the algorithm generation implementation, will be applied in the new data, which is like the vaccine generation process, The vaccine is facilitated by the generation of the disease, which in turn increases the effectiveness of the vaccine by interacting with the new disease, which is not a constant learning process.

Before the disease wants to be the training set, the vaccine wants to be the algorithm, and the new disease is the new input, the vaccine potency enhancement process is the algorithm learning process

In fact, machine learning can also be seen as a black box, a software process, with input and output

Continuous improvement of self in the process of input and output

And these knn, decision Tree, Bayesian, SVM, regression, are some rules and rules, to constrain and generate these algorithms.

Non-supervision and oversight

Non-supervised: classification and regression

Categories: KNN, Decision Tree, Bayesian, svm,adboost,logistics regression (results are discrete)

Regression: Linear regression (the result is continuous)

Supervision: Cluster recommendation

So machine learning is essentially some algorithm, but these algorithms are different from the general algorithm, their uniqueness is that their input is changing

------------------2016.4.25 written at Huazhong University of Science and Dong Shishe 318

Machine learning Personal Insights----late-night summary