TECHNICAL FORUM
Smart Sharing | [Lecture Series on Artificial Intelligence Methods and Applications] Lecture 2- Machine Learning and Neural Networks
September 10, 2021
Author: Deer Smart
View: 318
Machine Learning and Neural Networks
Artificial intelligence is a strategic technology that leads the future. Understanding the development process and current situation of the field of artificial intelligence, understanding the basic principles, applicable scenarios, and algorithm methods of machine learning, neural networks, and deep learning,
Through course learning, one can combine their own work with artificial intelligence thinking to think and solve practical problems.
A series of lectures on artificial intelligence methods and applications (which will be gradually shared with everyone in four chapters: the first, second, third, and fourth). The speaker is Dr. and postdoctoral researcher Lv Zengwei from Hefei University of Technology, who will provide a comprehensive presentation for everyone
Understanding the pure sharing of artificial intelligence is worth learning and collecting (thanks: due to the reproduction or adaptation of reference sources, the copyright belongs to the original author, and we would like to express our gratitude to the original author!). This is the second lecture - Machine Learning and Neural Networks.
1. Principles of Machine Learning
——Overview
Machine learning is a general term for a class of algorithms that attempt to extract implicit patterns from a large amount of historical data and use them for prediction or classification.
More specifically, machine learning can be seen as finding a function where the input is sample data and the output is the expected result. However, this function is too complex to express formally.
It should be noted that the goal of machine learning is to make the learned functions well applicable to "new samples", rather than just performing well on training samples.
Machine learning ≈ finding a function
1. Choosing a suitable model usually depends on the actual problem, and for different problems and tasks, it is necessary to choose the appropriate model, which is a set of functions.
2. To judge the quality of a function, it is necessary to establish a measurement standard, which is commonly known as the loss function.
3. Finding the "best" function and how to quickly find the "best" one among many functions is the biggest challenge, and achieving both speed and accuracy is often not an easy task.
After learning the "best" function, it is necessary to test it on a new sample. Only when it performs well on the new sample can it be considered a "good" function.
The No Free Lunch theorem (NFL), which means there is no gain without effort, is used to compare the relationship between two optimization algorithms,
How to determine if one algorithm is better than another. This theorem implies that no other algorithm can be better than linear enumeration or pure random search algorithms in search space.
——Terminology
Randomly select mango samples from the market (training data) and list all the characteristics of each mango, such as color, size, shape, origin, brand
And mango quality (output variables): sweet, juicy, and mature. Design a learning algorithm to learn the correlation model between mango features and output variables.
The next time you buy mangoes from the market, you can use the characteristics of the mangoes (test data)
Artificial intelligence is a strategic technology that leads the future. Understanding the development process and current situation of the field of artificial intelligence, understanding the basic principles, applicable scenarios, and algorithm methods of machine learning, neural networks, and deep learning,
Through course learning, one can combine their own work with artificial intelligence thinking to think and solve practical problems.
A series of lectures on artificial intelligence methods and applications (which will be gradually shared with everyone in four chapters: the first, second, third, and fourth). The speaker is Dr. and postdoctoral researcher Lv Zengwei from Hefei University of Technology, who will provide a comprehensive presentation for everyone
Understanding the pure sharing of artificial intelligence is worth learning and collecting (thanks: due to the reproduction or adaptation of reference sources, the copyright belongs to the original author, and we would like to express our gratitude to the original author!). This is the second lecture - Machine Learning and Neural Networks.
1. Principles of Machine Learning
——Overview
Machine learning is a general term for a class of algorithms that attempt to extract implicit patterns from a large amount of historical data and use them for prediction or classification.
More specifically, machine learning can be seen as finding a function where the input is sample data and the output is the expected result. However, this function is too complex to express formally.
It should be noted that the goal of machine learning is to make the learned functions well applicable to "new samples", rather than just performing well on training samples.
Machine learning ≈ finding a function
1. Choosing a suitable model usually depends on the actual problem, and for different problems and tasks, it is necessary to choose the appropriate model, which is a set of functions.
2. To judge the quality of a function, it is necessary to establish a measurement standard, which is commonly known as the loss function.
3. Finding the "best" function and how to quickly find the "best" one among many functions is the biggest challenge, and achieving both speed and accuracy is often not an easy task.
After learning the "best" function, it is necessary to test it on a new sample. Only when it performs well on the new sample can it be considered a "good" function.
The No Free Lunch theorem (NFL), which means there is no gain without effort, is used to compare the relationship between two optimization algorithms,
How to determine if one algorithm is better than another. This theorem implies that no other algorithm can be better than linear enumeration or pure random search algorithms in search space.
——Terminology
Randomly select mango samples from the market (training data) and list all the characteristics of each mango, such as color, size, shape, origin, brand
And mango quality (output variables): sweet, juicy, and mature. Design a learning algorithm to learn the correlation model between mango features and output variables.
The next time you buy mangoes from the market, you can use the characteristics of the mangoes (test data)
