1. Artificial Intelligence and Machine Learning
So that the public is often a confusing question is: artificial intelligence (AI), machine learning (ML), between deep learning (DL), machine vision (CV) as well as natural language processing (NLP) What is the relationship?
From the perspective of science Roughly speaking, covers over all the others AI [1], and a sub-machine learning of artificial intelligence, and the depth is a learning class of machine learning. As for machine vision and natural language processing, they are two specific applications in the field of artificial intelligence, and often use deep learning.
Figure 1. The relationship between artificial intelligence and related concepts
2. What is Machine Learning?
The more simple concept is actually more difficult to explain. For example, it was mentioned machine learning problem is in fact a 'optimization problems ", some people think that machine learning is a" programming concepts, "also suggested that at this stage of machine learning is" statistical inference. "
From a different perspective, these statements are reasonable. I personally like Tom Mitchell defined [1] for the "learning task" in:
Each machine learning can be accurately defined as: 1) task T; 2) training process E; 3) performance model P. The learning process can be disassembled as "In order to achieve the task T," we "Through the training process E" gradually "improve the performance of P" a process.
For example, we want to do a model to determine a picture of a cat or dog (task T). In order to improve the accuracy of the model (model performance P), we continue to model provides images allowed to learn the difference between cats and dogs (training process E). In this learning process, we get the final model is the product of machine learning, and the training process is the learning process. "Machine learning" and "human learning" is the analogy can do. Just apply the formula: In order to get a high score (task T) in the entrance, Wang day to do 10 sets of simulation questions (training E), and continue to participate in the test mode to detect their error rate (assessment P).
But in addition to the similarity outside, machines and people