Expert systems and Machine learning……
In the current generation if we ask what is the most eye catchy phrase that one should have heard,I am pretty sure the answer in a chorus would come out to be MACHINE LEARNING.
So without wasting any time lets get our hands dirty and dive deep into this world of Machine learning which can help us to get our dream job that most of you reading this article would aspire of.
Expert Systems
In AI, an expert system is a computer system that emulates the decision-making ability of a human expert. Expert systems are designed to solve complex problems by reasoning through bodies of knowledge, represented mainly as if then rather than through conventional procedural code.
In layman terms whatever rules are there inside a persons head ,we translate them to a program and let the machine decide the final output. This program is generally a combination of if-else conditions.
Applications of Expert systems in AI :
These Expert systems find extensive applications in
- Rocket Launching : Combination of if-else conditions decide various events leading to launch of a rocket
- Loan: Used by banks to decide whom to issue a loan or not on basis of various factors
- Speech Synthesizer systems.
Limitations of Expert Systems :
- In the current data is being produced at a massive rate,it becomes difficult for any human being to come up with the if-else rules and interpret them
- Even if these rules are formed,they would be quite complex
- Rules that are formed are sometimes inexpressible (for example consider the task of hiring a person,the recruiter wants to hire a particular person because he saw honesty in his eyes).such qualities are inexpressible with numbers
- Sometimes rules can be unknown.(for instance task of predicting ebola) which was a new disease.
These limitations of expert system give rise to Machine Learning,wherin instead of a human coming up with these rules we give all the input data to a machine and let it figure it out what is the best possible decision for that particular task.
Reason for success of ML
Success of ML can be attributed to these 3 factors
- Abundant data : In times to come the pace at which data is being produced will increase exponentially so it is bound that machine learning algorithms will outprform any human expert
- Democratization: with the source code being available online anyone can get into machine learning and reap its benefits and make a career out of it.
- Devices: The modern day devices are relatively cheap and with cloud services being offered at reasonable prices it becomes very easy to deploy ML related models.
Different roles in Machine learning:
- Data engineers : These people collect ,create data,sometimes responsible for data preprocessing and cleaning it.
- ML engineers: These are responsible for creating solution and applying various algorithms,They decide which algorithm to use and which not to under a given condition.
- ML researcher: These are responsible for coming up with optimized versions of a particular algorithm.
To draw an anology,lets consider a case where we want to build a car. the researchers would try to build a car that can be run everywhere . The ML engineers will help us how to drive these cars which car to choose under a different situation and lastly data engineers will decide about destinations where we want to take our car.