I. Three Major Machine Learning Algorithms
1. Decision Tree Algorithm
The method first divides the samples into different subsets, then recursively divides them until each subset contains samples of the same type. Testing begins at the root node, proceeds to the subtree, and then to the leaf node to arrive at the predicted category. This method is characterized by its simple structure and high data processing efficiency.
2. Naive Bayes Algorithm
The Naive Bayes algorithm is a classification algorithm. It's not a single algorithm, but a family of algorithms, all sharing the common principle that each feature being classified is independent of the values of any other feature. The Naive Bayes classifier assumes that each of these "features" contributes a probability independently, regardless of any correlation between features. However, features are not always independent, which is often considered a drawback of the Naive Bayes algorithm. In short, the Naive Bayes algorithm allows us to predict a class using a set of features given probabilities. Compared to other common classification methods, the Naive Bayes algorithm requires very little training. The only work that must be done before making predictions is finding the parameters of the individual probability distribution of the features, which can usually be done quickly and deterministically. This means that the Naive Bayes classifier can perform well even with high-dimensional or large amounts of data.
3. Support Vector Machine Algorithm
The basic idea can be summarized as follows: First, a transformation is used to increase the dimensionality of the space; this transformation is non-linear. Then, the optimal linear classification surface is obtained in the new complex space. The classification function obtained in this way is formally similar to a neural network algorithm. Support Vector Machines (SVMs) are a representative algorithm in the field of statistical learning, but their approach differs significantly from traditional methods. By increasing the dimensionality of the input space, the problem is simplified, reducing it to a classically linearly separable problem. SVMs are applied to various classification problems such as spam detection and face recognition.
II. What is the connection between AI, machine learning, and deep learning?
Machine learning is a method for achieving artificial intelligence. The most basic approach of machine learning is to use algorithms to analyze data, learn from it, and then make decisions and predictions about real-world events. Unlike traditional software programs coded to solve specific tasks, machine learning is "trained" with massive amounts of data, learning how to perform tasks through various algorithms. Traditional machine learning algorithms include decision trees, clustering, and Bayesian classification. From the perspective of learning methods, it can be divided into supervised learning, unsupervised learning, semi-supervised learning, ensemble learning, deep learning, and reinforcement learning.
Deep learning is a technique for implementing machine learning. Initially, deep learning was a learning process that used deep neural networks to solve feature representation problems (deep neural networks can be roughly understood as neural network structures containing multiple hidden layers). Deep learning was not originally an independent learning method; it also used supervised and unsupervised learning methods to train deep neural networks. However, due to the rapid development of this field in recent years, some unique learning techniques have been proposed (such as residual networks), leading more and more people to view it as a standalone learning method.
In summary, machine learning is a method for achieving artificial intelligence, while deep learning is a technology for achieving machine learning. Furthermore, it can be seen that the artificial intelligence currently touted by robot manufacturers worldwide primarily focuses on weak AI. This involves collecting surrounding information by adding various types of sensors such as visual sensors, force sensors, and LiDAR, supporting mainstream deep learning frameworks, and utilizing intelligent algorithm libraries to improve the robot's performance in tasks such as interaction, perception, recognition, classification, and decision-making. While there is still a significant gap between now and achieving strong AI, continuous advancements in software and hardware technologies, along with the accumulation of basic research, may allow us to achieve strong AI in the near future.
