Shulou(Shulou.com)06/01 Report--

This article introduces the relevant knowledge of "what is the generating model and distinguishing model in machine learning". In the operation of actual cases, many people will encounter such a dilemma. Next, let the editor lead you to learn how to deal with these situations. I hope you can read it carefully and be able to achieve something!

What are the two models?

Let's move on to these two concepts in a few words:

1. Machine learning is divided into supervised machine learning and unsupervised machine learning

2. Supervised machine learning is to train the classifier by knowing the category of the training set data, and unsupervised machine learning is to train the classifier without knowing the category of the training set.

3. Therefore, supervised machine learning can be abstracted as a classified task, while unsupervised machine learning basically completes clustering.

4. In supervised machine learning, we can summarize that we can train a model through a lot of marked data, and then use this to predict the Y of the input X. There are generally two kinds of models:

Decision function: Yampf (X)

Conditional probability distribution: P (Y | X)

5. According to the method of obtaining these two models by learning data, we can be divided into discrimination method and generation method.

6. Formal introduction of the concept

Discriminant method: the discriminant model is used to learn the decision function Y (X) or conditional probability distribution P (Y | X) directly from the data. The discriminant method is concerned with what kind of output Y should be predicted for a given input X.

The prediction model obtained by learning the decision function Y (X) or conditional probability distribution P (Y | X) directly from the data is the discriminant model.

Generation method: the joint probability distribution P (XMagi Y) is learned from the data, and then the probability distribution P (Y | X) is calculated from P (Y | X) = P (X) / P (X) as the prediction model. This method expresses the generation relationship between the given input X and the generated output Y.

The prediction model of P (Y | X) is to generate the model.

Examples of two models

Generation model: naive Bayesian (Naive Bayes), hidden Markov (EM algorithm)

Discriminant models: K nearest neighbor method, perceptron, decision tree, logical regression, linear regression, maximum entropy model, support vector machine (SVM), lifting method, conditional random field (CRF), neural network

Contrast

1. The generation model can restore the joint probability distribution (restore the similarity of the data itself), but the discriminant method can not.

2. The learning convergence speed of the generation method is faster, and when the sample size increases, the learned model can converge faster to the real model.

3. When there are hidden variables, the generation method can still be used to learn, but the discrimination method cannot be used at this time.

4. Discriminant learning can not reflect the characteristics of the training data itself, but it looks for the optimal classification surface between different categories and reflects the differences between different kinds of data. Facing the prediction directly, the learning accuracy is often higher. Because learning P (Y | X) or Yamazf (X) directly, the learning can be simplified.

5. To put it simply, the generation model is to find rules from a large number of data, which belongs to statistical learning, while the discriminant model only cares about the differences of different types of data and uses differences to classify them.

The discriminant model is to predict whether your biological father will be taller than 180 under known conditions. If your biological father is higher than 180, then your chances of being higher than 180 will increase. But the probability distribution of height around the world has not changed for the time being. (for the sake of scientific rigor, the father here has been changed to a biological father rather than a father.)

The generated model randomly gives you the tallest height in adulthood. The probability of a height exceeding 180 can only be determined by the frequency distribution of the height of all adults around the world.

This is the end of the content of "what is the generating model and discriminating model in machine learning". Thank you for your reading. If you want to know more about the industry, you can follow the website, the editor will output more high-quality practical articles for you!

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.