Expert Systems & Machine Learning — PadhAI DL Course

Source: Deep Learning on Medium


Photo by Hal Gatewood on Unsplash

Expert system is a computer system that emulates the decision-making ability of a human expert

Expert system are also known as Rule Based Systems. How does a human being make a decision? They look at inputs (features) and then based on previous experience, they decide the output / result.

An example of a doctor’s past experience with previous patients

To develop computers to make decisions, we can represent the above data by converting green to 1, red to 0.

What is the semantics of decision making?

  1. Features (Inputs)
  2. Rules (Combines features into some rule base)

Now we can make computers predict by converting the rules to simple if and else statements

Some limitations of Expert Systems:

  1. Tedious when there is a lot of data
  2. The rules can be unknown (E.g: Symptoms for Ebola)
  3. The relationship between different features can be highly complex and hence the rules can be too difficult to figure out
  4. The rules may be inexpressible (E.g: Emotions)

Machine Learning

In machine learning, we give some data and we ask the machine to find a model that best explains the relation between input and output.

Goal is finding a function that maps input to output by learning to find the correct set of parameters.

Why is ML very successful now? — 3D’s
1. Data
2. Democratization
3. Devices

  1. There is abundant data now.
  2. We have access to wide variety of algorithms and functions and different people have made their effort and code public.
  3. We now have fast and cheap cloud computing solutions.

The key difference between Expert Systems and ML is:
In Expert System we have a rule base. And in ML, we have a family of functions and algorithms that allow us to learn the parameters.

All images courtesy of PadhAI One Fourth Labs