This article is the learning excerpt and notes recorded by the blogger for personal study, research or appreciation when he is studying in artificial intelligence and other fields, and based on the blogger’s understanding of artificial intelligence and other fields. I will correct it immediately after pointing it out, and I hope for your understanding.
Articles are categorized in the study excerpts and notes column :
Study excerpts and notes (2) --- " The Mystery of Artificial Intelligence: The Schools of Machine Learning"
The Mystery of Artificial Intelligence: The Schools of Machine Learning
Source of original text/paper:
Topic: " The Mystery of Artificial Intelligence: Huashan Discusses Swords among the Schools of Machine Learning, Who Can Survive the Swordsmanship? "
Author: Zhu Liming
Time: 2022-09-28 07:29
Source: Published in Beijing.
Turing Award winner and head of Meta artificial intelligence, Yann LeCun (Yann LeCun) proposed a famous cake metaphor in 2016: "If intelligence is a piece of cake, the main body is self-supervised learning, the icing on the surface is supervised learning, and the dotted cherry is reinforcement learning”
machine learning :
A method of realizing artificial intelligence that continuously improves the performance of a certain task from data through a certain algorithm model,
Taking it a step further, deep learning is a machine learning method that uses neural network models to learn from massive amounts of data.
Supervised Learning
The most commonly used method of machine learning, which uses data sets labeled by experts to train models.
Similar to: Literacy by pictures
Unsupervised Learning
No labeling of the data is required. Unsupervised learning can learn the laws contained in the data itself or learn its internal representation, so new data can be generated. Generally less accurate than supervised learning
Self-Supervised Learning
Use massive unlabeled data to automatically label the data through the laws contained in the data itself, so as to transform unsupervised learning into supervised learning.
Semi-supervised learning
Combined use of supervised learning and unsupervised learning machine learning methods. It can use a small amount of labeled data and a large amount of unlabeled data for model training, so it is also called weak-supervised learning.
The difference between the above machine learning methods mainly lies in the number of labels in the training data set .
Supervised learning is full labeling, semi-supervised learning is a small amount of labeling, unsupervised learning does not require labeling at all, and self-supervised learning is to find out potential labels from unlabeled data.
Deep Learning
Reinforcement Learning
Instead of learning from a static dataset, study how the agent makes optimal continuous decisions in the environment (gets the maximum reward)
Reinforcement learning is similar to how animals and humans optimize behavioral patterns by rewarding and punishing interactions with the environment .
Deep Reinforcement Learning
The combination of reinforcement learning and deep learning has made great progress in recent years.
If the article is inappropriate and incorrect, please understand and point it out. Since some texts and pictures come from the Internet, the real source cannot be verified. If there is any dispute, please contact the blogger to delete it. If you have any errors, questions or infringements, please leave a comment to contact the author, or follow the VX public account: Rain21321, and contact the author.