[Artificial Intelligence] acquaintance

artificial intelligence

Excerpt from "deep learning Pytorch".

definition

AI (Artificial Intelligence), also known as machine intelligence, refers to a system of artificial intelligence created by the performance. The so-called smart, that value can observe the surroundings and to take action accordingly has the purpose.

classification

The concept of artificial intelligence is very broad, and now it is an example of artificial intelligence will be divided into three categories:

• Weak AI (Artificial Narrow Intelligence, ANI)

AI AI is a weak single aspect of good, such as AlphaGo.

• Strong AI (Artificial General Intelligence, AGI)

Achieve human-level artificial intelligence. Strong AI means every respect and humanity par artificial intelligence, human beings capable of living brain, which are competent. The creation of strong AI is much more difficult than creating artificial intelligence is weak, there is not. Linda Gottfredson when the professor intelligence defined as "a broad mental ability to think, plan, problem solving, abstract thinking, understand complex ideas, learn quickly and learn from experience and other operations." Strong AI during these operations It should be handy and humans.

• Ultra AI (Artificial Super Intelligence, ASI)

Oxford philosopher, well-known idea of artificial intelligence and Nick Bostrom the super-smart defined as "the most intelligent than almost all areas of the human brain are a lot smarter, including scientific innovation, through knowledge and social skills." Super AI can be in all aspects a little stronger than humans, it may be stronger in all aspects than human trillion times.

Data mining, machine learning and deep learning

Data Mining

In short, data mining is to find useful information in large databases, and make the process of analysis, that is, they say, KDD (Knowledge Discovery in Database) .
A process of data, is to start from the input data, preprocessing of the data, including feature selection, normalized, reduced dimensions, the data to enhance the like, and then analyzed and data mining, and then processed, for example pattern recognition, visualization Finally, the entire process of formation of the information available.
So that data mining is only a concept, that is, from data mining into meaningful information, looking for characteristics between the data from large amounts of data.

Machine Learning

Machine Learning regarded as a way to achieve artificial intelligence, data mining, and it has some similarities, but also more than one field of cross-disciplinary, involving probability theory, statistics, approximation theory, convex analysis, computational complexity theory and other subjects . Different from each other to find the data characteristics from among a large data mining, machine learning, pay more attention to design algorithms, so that the computer can automatically from the data in the "learning" rule, and the unknown data to predict the use of the law. Because learning algorithm involves a large number of statistical theory, particularly in close contact with statistical inference, it is also known as statistical learning methods.

classification

Machine learning can be divided into the following five categories:

1. Supervised learning: from a given training data set out a learning function, when the arrival of new data, the results can be predicted based on this function. Supervised learning the training set input and output requirements, it can be said characteristics and objectives. The goal of the training set is marked by the people. Common supervised learning algorithms including regression and classification.

2. Unsupervised learning: unsupervised learning and supervised learning compared to the results of the training set no artificial label, a common unsupervised learning clustering algorithm.

3. Semi-supervised learning: methods between supervised learning and unsupervised learning between.

4. Transfer learning: the trained model parameters have been migrated to the new model to help the new model training data set.

5. Enhanced Learning: learning by observing the surroundings. Each action will have an impact on the environment, learning objects to make judgments based on the feedback observed the surroundings.

Machine learning algorithms

Traditional machine learning algorithms are the following: linear regression, logistic regression model, k- near algorithms, decision trees, random forest, support vector machines, artificial neural networks, EM algorithm, probabilistic graphical model .

Depth study

Depth study of the most basic version of the artificial neural network, which is a branch of machine learning, which attempts to simulate the human brain, characterized by automatically extracting data more complex structures.

Deep learning structure

With the development of neural networks, the more popular network structure were: the depth of the neural network (DNN), convolutional neural network (CNN), the cycle of recurrent neural network (RNN), generating confrontation Network (GAN) , and so on.