Which way to MACHINE LEARNING ?
Anand Khandekar
ArcticTurnFoundation | Validus Analytics | ValidusEduTech | IoT | EDGE AI | Sustainability | Research | SDG
Of late I received a call where the conversation starts like this ," hey prof ! I am very interested in getting into Machine Learning. The dilemma is : where to begin"
More and more mid and higher managers from finance, banking, insurance and manufacturing sectors are contemplating the need to delve into understanding the implications of ML. There seems to be a kind of a pressure being sensed by one and all to keep up with the change in scenarios.
Enterprises, across sectors have already implemented OCR and RPA applications in an attempt to streamline their processes. Automation seems to be the buzzword doing rounds in every meeting and presentation. The bombardment of terms like Machine Learning Artificial Intelligence still continues to happen.
Indian Industries seem to have graduated into phase 2 as I like to call it. The domain experts in organisations are feeling the need to understand how the ML and AI applications work. In fact even with the advent of DEEP NEURAL NETWORKS ad our friendly NETFLIX making recommendations on the basis of similar such advanced technologies, most of the lower pyramid of industries still can be very effectively serviced using basic ML fundamentals.
In the words of Andrew Ng, co-founder at COURSERA, " most of the insights needed today can still be obtained by implementing ML algorithms".
For the beginner, in order to start, I recommend jumping into the sea by firstly understanding a few implications. What is it that you wish to use the ML tools for ? Where to begin depends on what insights that you wish to derive. Get clarity on this aspect. Listed below is a broad classification of the nature of insights you may seek :
- FORECASTING : e.g. Sales , Energy demand prediction
- DETECTING ANOMALIES : e.g. Intruders, virus mutations etc
- CLASSIFICATION : e.g. images identification, cancer detection form MRI, spam
- SUMMARIZE : e.g. sentiment analysis
- OPTICAL CHARACTER RECOGNITION : e.g. document analysis
The list could go on forever. For the BEGINNER my suggestion is to understand his/her domain or the nature of work they are currently trying to AUTOMATE. Ask questions like : what is the most pressing need of such an implementation? Is your organisation in a position to invest on getting Proof Of Concept (POC) done? By choosing your industry/company specific project, you ensure that related internal workforce benefit immediately and understand the value.
Consider collaborating with external partners initially to attach expertise. This will help you understand the intricacies and will lead to you taking the RIGHT PATH in the vast network of ML applications.
"However, in order to truly rise to the challenge of formulating a winning workforce strategy for the Fourth Industrial Revolution, businesses will need to recognize human capital investment as an asset rather than a liability." says The Future of Jobs Report 2018 published by World Economic Forum.
Runner | Reader| Solution Architect | Power Platform | Blue Prism | Intelligent Automation | RPA | Digital transformation
5 年Excellent piece, on back of our discussion Prof. Awaiting for next in line now....
Building and Scaling Businesses on Marketplaces | RetailEZ | Ex Trell | Ex Hopscotch
5 年It is a great article, Professor. Automation and Machine Learning are slowly taking over the most important part of the industry, and sadly i believe that our syllabus doesn't contain enough material. Do you think we could set up an add on in our college to facilitate guiding students towards this??
Empowering Leaders to Break Barriers | Executive Coach | Leadership Mentor | 30+ Years of Transforming Careers and Teams | Founder, Nirvedha Executive Coaching Solutions | Certified Board Director
5 年Nice article Anand