Machine Learning seems to be the future in any services with a large number of customers. Companies/organizations may be in a big trouble if they do not learn about customers from their big collected data using machine learning methods. However, to this end, they may need 3 in 1 employers: strong software engineering skills, comprehensive knowledge about machine learning methods, and deeply understanding their data as well as their main targets.
Both Google and Amazon provide platforms for machine learning systems with affordable prices for nearly everyone. Any small or medium companies can set up a systems to upload data and apply basic methods to learn or mine knowledge or unknown information or patterns from their data. However, machine learning is not a game for conventional software engineers. Even the best software engineers (as Google, Amazon or Facebook employees) will struggle to understand models or principle behinds machine learning methods, let a lone interpret the meaning of results. Without a comprehensive knowledge and experiences in Machine Learning, conventional software engineers will not be more helpful than monkey coders.
Another issues with machine learning applications is that it is impossible to jump into the markets to find software today and apply to any data tomorrow. Each companies or organizations need to invest into some one to understand their problems, their data, and their targets. Then the person will seek for suitable available models or methods, implement the methods, and interpret correctly the results. To this end, he or she must have a strong software engineering skills and comprehensive knowledge of machine learning. He or she must be able to convert finding into action plans to get the target. In the other words, any companies or organizations need to build a group of 3 in 1 with high experience in software development if they want to apply machine learning for their business.
Although the move of Google and Amazon can make the whole markets moving forward to machine learning applications to real life data, the services include very few basic methods (e.g., linear logistics, neural network). The methods would provide a very first steps in understanding data as well as patterns, they are far way from real-useful applications. I bet that systems or methods used inside Google would be much more comprehensive than the systems provided for published users. I bet Amazon would do the same. Thus, companies or organizations may need to build customized systems for themselves.
In short: Machine Learning will be the future of business. It is hard to image any companies or organizations would survive in the next 5 years without using machine learning for their business. However, to use machine learning is not a simple game but an expensive investment if they want to fulfill their dream.
Author: Quang Si Le