If you are one among the many, who is keen to understand what is machine learning and why you need to understand or use this tech, then read this further.
I will try and breakdown the differences between automation and machine learning based on some of the fundamental principles of how computers work. Both are going to be run on processing computers, microprocessors or controllers.
The fundamental difference arises whether the decision making system is simple or complex. A simple system will require converting the idea or logic into a sequence of decision making steps and then coded in to a program. The dependency to code here will be the rules. Eg; If weight is less than 55Kg selected. If Blood pressure is greater than 90/130 selected etc.
Defining a conditional logic vs deriving the logic using data.
You may think now, it should be similar for a complex system also right, but if the number of decision making required is going to be enormous or just simply because it is not possible to explain the individual decision making itself. But the end outcome can be explained, then the dependency will be on the data that is generated. Lets understand through a example; There is a HR selecting a candidate based on a personal interview performance. The outcome is either select or reject. But to explain the rules of selection based on personal interview is not easy. It is more of subjective and differs from person to person. Can we define some criteria? Like how the candidate is evaluated for a customer executive,
The answers should be positive, without any errors in usage of the language.
The candidate should look confident and answer all questions without any struggle.
A human being is naturally good at understanding emotions and language even though the inference is based on huge data i.e audio visual communication in which they are involved. The objective now will be to make the computer understand this data in each of the case whether select or reject.
This is one of the fundamental aspects of difference between a automation and machine learning.
A machine learning system is reliant on data to derive its rules which would otherwise be not possible.
Conclusion
There must be a goal defined for a machine learning system also, like defining the objective of an automation system.
In any profession or business there is increasing reliant on automation and data is a byproduct. Can this data be put to use?
Some points to think,
- Can you increase the effectiveness of the system
- Can the customer experience be improved, etc.
The system could be engineering, medical, financial or anything. To be able to look at the opportunities for yourself in your relevant area a mix of both the domain knowledge and also the general machine learning decision making tasks one should be aware of.