Improving the Machine Learning algorithm - Getting an upstart

Artificial Intelligence is the field of the future. It is going to pervade all spheres of human activity. The earlier one gains access to the terms and tools of using AI and Machine Learning, the better one can remain in the forefront.

Machine learning is an application of artificial intelligence (AI) that provides systems the ability to automatically learn and improve from experience without being explicitly programmed. Machine learning focuses on the development of computer programs that can access data and use it learn for themselves. - expertsystem.com

Machine Learning is completely algorithm, data and feedback based. The output of the Machine Learning system is the output of the interaction of data and feedback with the algorithm. 

One of the best examples of Machine Learning is the Google Maps. When we look at how to improve the Machine Learning algorithm that improves the predictive behavior of the system, we need to look at the following steps and analyses

1. the driving routes, 
2. number of vehicles using the Google Maps software, 
3. alternate routes available
4. how fast one can collect this data through the communications network, 
5. how fast it can be processed, 
6. how fast can the feedback be collected and 
7. how effectively can this feedback be recursively used to improve the algorithm

This is just one example of Artificial Intelligence /  Machine Learning application in the modern day life.  Analysing Radiographic images is another area where AI is already finding use.

So the best way to improve the machine learning is to find ways 

1. to improve the algorithm based on which the Machine learning happens
2. getting access to reliable, accurate, disparate and varied source of training data and
3. gaining access to faster and effective feedback loop based on which the algorithm improves further
4. provide best quality output results that are effective
   
5. gain access to competitive advantage in sourcing data, and feedback

In Machine Learning we cannot deliver less quality results like in products, it can only be less effective output or prediction. Since Machine Learning is self learning and adaptive, if we give it access to large amounts of quality data and feedback, naturally the output predictions also will be of high quality.

Gaining the Competitive edge is vital in the Machine Learning area as the saying goes, the winner takes it all. 

Each player distinguish himself and his AI services by 

1. identifying sources that can give data and feedback that has not been accessed, 
2. identifying sources of data and feedback that are slow and hence has not been accessed, 
3. identifying sources of data and feedback that are not in a readable form 

The player who has access to large quantities of data and feedback and having an efficient algorithm that adapts to the changing times,. will finally get the upper hand in the coming years. 

The article by Ajay Agarwal et al, "How to win with Machine Learning", HBR, Oct 2020 (click here) gave a good insight into writing this article. 

George..