Once beginners begin to try to understand this problem, they often find well-meaning but frustrating suggestions, as follows:
You need to be proficient in mathematics. The following is the list:
-Calculus
differential equation
mathematical statistics
-Optimization
-Algorithm analysis
...
A reply like this is enough to scare anyone, even those with a certain mathematical foundation.
I guess many beginners will be intimidated by such advice, but in fact they need less math knowledge than you think (at least less than others tell you). If you are interested in becoming a machine learning practitioner, you don't need a lot of advanced mathematics knowledge to start.
But if there is no threshold, it is not. In fact, even without a high understanding of calculus and linear algebra, there are other thresholds.
Mathematics is not the main premise of machine learning.
If you are a beginner, and the goal is to deal with the problems of industry or enterprise, then mathematics is not the main prerequisite for machine learning.
So far, most of the advice you have heard about machine learning comes from experts engaged in data science in academic fields.
In the academic field, you are often encouraged to do academic research and write reports. When your research field is machine learning, then you really need to have a deep understanding of the statistical and mathematical basis of machine learning.
In the industrial field, in most cases, the main pursuit is not invention (making wheels) and writing reports. Whether the enterprise really pursues to create business value. Many times, especially in the early days, it is enough for you to use "ready-made" tools. At this time, you will find that the requirements of these tools for mathematics are not as high as you think.
"Off-the-shelf" tools don't need high mathematics.
Almost all common machine learning libraries and tools will handle difficult math problems for you, which means that you don't need to know linear algebra and calculus to engage in machine learning.
Again, modern statistics and machine learning software can help you deal with many math problems.
For beginners, the mathematical knowledge involved in machine learning is as deep as the sea, and it is not necessary or necessary to understand the mathematical knowledge in the deep sea field.
Of course, these tools cannot do everything for you. You still need practice to master these tools.
If you want to start learning machines, the really necessary skill you need to learn is data analysis.
For beginners (whether you are a software engineer or a practitioner from other fields), you don't need to know much about calculus, linear algebra or any other university-level mathematics to complete these tasks.
However, the ability of data analysis is essential. Data analysis is the first skill you need to complete your work, and it is the necessary ability that beginners of machine learning really need.
Mathematics is very important, but it is not suitable for beginners.
Mathematics is very important, especially in some cases.
First of all, if you are doing machine learning research in the academic field, mathematics is very important; Secondly, in the industrial field, mathematics is also very important for a few senior data analysts/data scientists. In particular, companies like Google and Facebook are at the forefront and are using cutting-edge tools in the field of machine learning. These people often use calculus, linear algebra and more advanced mathematics in their work.
Beginners also need mathematics to learn machine learning. Starting to learn machine learning requires at least entry-level undergraduate basic mathematics skills. You also need to know basic statistical knowledge, such as average, standard deviation, difference and so on.