Q1 How to classify machine learning

According to the task type, it can be divided into:

  • “Regression models” : predicting tomorrow’s share price, for example.
  • “Classification model” : Divide samples into two or more categories.
  • “Structured learning models” : Outputs are not vectors but other “structures.”

According to learning theory, it can be divided into:

  • “Supervised learning” : learning samples “all” have labels, training the network to get an optimal model.
  • “Unsupervised learning” : Samples of training are “all” unlabeled, such as “cluster samples.”
  • “Semi-supervised learning” : Training samples are labeled “partially.”
  • Reinforcement learning: A mode of training in which an agent interacts with the environment to receive “rewards”. The environment does not judge whether it is correct, but continuously “evaluates” the agent’s behavior with feedback signals.
  • “Transfer learning” : using the trained model to learn new samples, mainly to solve the problem of too few samples in the problem.

Q2 What are discriminant and generative models

“Discriminant method“

Directly by the data to learn the decision function Y = f (X), or by the conditional distribution probability P (Y | X) as a predictive model for discriminant model.

The common discriminant models include Linear regression, Boosting, SVM, decision tree, perceptron, Linear discriminant analysis (LDA), Logistic regression and other algorithms.

“Generating methods“

By learning x and y data on the number of the joint probability density distribution function P (y, x), and then by the bayesian formula and conditional probability distribution P (y | x) as a prediction model for the generated model.

Common generation models include naive Bayes, hidden Markov model, Gaussian mixture model, document topic generation model (LDA) and so on.

“For example,“

Decide whether an animal is an elephant or a cat. Remember that an elephant has a long trunk to tell which is the elephant. Draw an elephant and a cat and compare them with the animal.

Hi meet?

❝

Wechat public Account:“They’re all code farmers“ (allmanong)

You will get:

All the interview questions about artificial intelligence“The majestic“! There is still“Mind mapping“Oh!

reply“121“121 books have been sorted out immediately“Python Learning Ebook“.

reply“89“Immediately available“The programmer“Epic level required reading single spit blood collation“Four dimensions“Series 89 books.

reply“167“Immediately available“Machine learning and Python“The history of the road to learning“Big data Technology book“From Beginner to Advanced Most complete book (66)

reply“18“Immediately available“The database“From the beginning to the advanced compulsory reading 18 technical books online disk collating e-books (rare edition)

reply“56“Get the 56 books I compiled immediately“Algorithms and data structures“book

In the future, there will be artificial intelligence postgraduate course notes and so on. We will make progress together!

❞