Big Data: Possibilities and challenges
At the 2016 edition of Mint’s Enterprise Tech Summit titled ‘Big Data analytics in an IoT world’, experts discuss potential and barriers to widespread IOT-Big Data value delivery
- Cracking India’s bankruptcy code
- RBI’s registry will help solve problem of credit shortage: iSpirt’s Sharad Sharma
- Fintech regulation at an inflection point: Shardul Amarchand’s Shilpa Mankar Ahluwalia
- Voice and AI biggest transformative tech, says EY’s Mahesh Makhija
- User consent could be central to data privacy law, says Tanuj Bhojwani of iSpirt
Big Data and The Internet of Things (IoT) are intimately connected: billions of Internet-connected ‘things’ will, by definition, generate massive amounts of data. Companies are using Big Data analytics for everything from driving growth, reducing cost, improving operational excellence, recruiting better people to completely transforming their business strategy. At the 2016 edition of Mint’s Enterprise Tech Summit titled ‘Big Data analytics in an IoT world’, a panel of experts discussed the potential and barriers to widespread IoT-Big Data value delivery, the lack of standards and concerns over security and privacy.
The panellists included Shriram Revankar, vice president and Fellow, Adobe Research Big Data Experience Lab; Srinivas Aravapalli, principal chief engineer and head of vehicle systems at Mahindra and Mahindra Ltd; Seemantini Godbole, vice president, digital technology, Target India; Vishal Dhupar, managing director, South Asia, NVIDIA Graphics; Arun Vinayak, Chief Product Officer and Founding Team member at Ather Energy; Vinay Jammu, technology leader in GE Global Research, and part of the Software Science Analytics Team at GE; Amandeep Sarna, regional director, IT, South Asia at Starwood Hotels and Resorts Worldwide, Inc; Uday Prabhu, general manager, Internet of Things, Robert Bosch Engineering and Business Solutions Ltd; Akilur Rahman, head, corporate research center, ABB India; Sanchit Gogia, chief analyst and CEO of Greyhound Research; and Jayanth Kolla, founder and partner at Convergence Catalyst.
The panels were moderated by Leslie D’Monte, national technology editor of Mint. Edited excerpts:
Do companies understand what Big Data is?
Gogia: If you spoke to clients two years ago, they would say we don’t understand Big Data and would want to pause and see. But now, the response is I understand Big Data, but what I understand of it is the marketing aspect of it, but I don’t understand all the use cases. Possibly, my industry hasn’t had so many use cases. Interestingly, some of the firms have a separate Big Data officer and a separate BI (business intelligence) officer. That is a recipe for disaster. What you are actually getting after is the same data that can be beneficial to your business. There are tonnes of firms that don’t really understand where the starting point is.
What about the jump in computing capacity?
Dhupar: This is a world where there is so much information. Look at 2015. There was a confluence of label data along with highly affordable parallel programming processors that started changing the fashion of work. What 2015 has done for us is not only going to make a change in computer science, but the computing model. Why do I say this? In 2015, Google and Microsoft did something. Firstly, Google demonstrated that machines can recognise images better than human beings. Microsoft went one step ahead and created a neural network. The previous network was created of 18 layers in AlexNet. They now created 180 layers. You could now feel what the computer could really do along with human beings which humans couldn’t.
If you have so much data, you should understand what can be done. If we talk about the entire paradigm, that is where deep learning is happening, especially the start-ups, you are talking about Big Data and IoT, about how can you make algorithm that will infinitely make your work shine.
Revenkar: Essentially the optimisations that we are going after are no longer necessary. Professor Kern from University of Penn demonstrates in the world of social networks how traditional algorithms that are considered to be perfect, no longer produce optimal answers, especially when social nodes are very large. We can now compute things very differently than what traditional computer science taught us. Deep Learning is one of the radical improvements.
How do you decide which data is useful and which is not?
Godbole: There is data everywhere. You can’t go after it all. I feel the opportunity is outsizing what we are doing currently, and that basically for a business, it comes down to: where do you see the most value for your gas?
I think the speed is becoming much faster because of the fact that everything is available now. You had to be in a position of privilege if you had to do it sometime back. You had to work for a big company, have licensed software, have access to storage and computing. Now, whether you are in a start-up or a Fortune 500 company, you could possibly do the same level of analysis in techniques such as deep learning.
What are some of the challenges of using Big Data?
Gogia: The first problem we face is the inability to bring business owners on to the table. CIOs (chief investment officers) and IT leaders struggle in making use cases. Second issue is, where do we start from?
Give us three highlights of the research you have done in consumer and industrial IoT?
Kolla: Big companies, industries and enterprises are used to growing incrementally and thinking incrementally, but when it comes to IoT and its capabilities, these companies are opening up to this paradigm shift. The second thing we are observing is collaboration and partnerships. Big companies are collaborating with younger companies start-ups, app developers, and within themselves too. The third trend is that these companies are opening up their legacy systems to adopt the new solutions that are coming from the IoT space. In India, for instance, Manipal Hospitals is using a start-up, Cardiac Design Labs’ ECG (electrocardiogram) machine and is open to replacing its big GE and Philips ECG machines.
But Big Data integration with legacy systems poses an important challenge, especially in the manufacturing sector...
Jammu: When we look at what we have today, we focus on cloud, that’s the newest technology, but at the same time all the previous technologies for the past 30 years continue to exist. What we need to do, is move them one piece at a time and we need to be able to build the ecosystem.
How does one look at return on investment (RoI) from IoT?
Rahman: When we talk about IoT, RoI is also in terms of reducing the cost of poor quality. In the IoT paradigm, we can proactively monitor things and take action so that we don’t end up in cost of poor quality. Secondly, it is also how you increase the productivity in terms of energy efficiency, safety environment, how do you utilise the machines, the assembly line, the workforce, reduce the cost of the material, reduce the scraps.
Prabhu: RoI is definitely something that needs to have connotation with respect to what are you trying to achieve. How do you ensure that products are getting out faster? How do you ensure that there is lesser wastage? So I think IoT is an expensive investment if you try to do a cluster-bomb across the production line. You need to look at areas which require urgent attention, the bottlenecks, clear the bottlenecks, you get immediate returns, take care of the low hanging fruits first and then go to bigger things.
But given that IoT is an imperative, is looking at RoI a fair proposition?
Kolla: It’s an eventuality, an evolution. But we’re still in the early stages, so looking at RoI at this stage would hinder innovation and potential to a significant extent.
How do you address these two very important concerns—security and privacy, when everything is connected?
Vinayak: The security levels on our systems are as complex to hack as any of your phones today, and I am pretty certain you have more critical data on your phone today than you will have on our system. Added to that, we do give you the option to disengage the location services so you don’t need to share your location with us. The rest of the data is really not so critical.
Sarna: There are two things at play here. One is the mindset of the consumer which is very important. Today, we do a lot of surveys and we ask our customers whether you are comfortable sharing that kind of data with us, provided we encrypt and store it. Surprisingly, more than 50% of our customers come back and tell us they are fine with that. Second thing is, there are no prevailing laws in India. But thankfully for us, being a US company, there is a governing body, it’s called PII (personally identifiable information), and we are mandated to encrypt whatever information we get from our guests and use it only at the discretion of the guest.
IoT, Machine Learning, Artificial Intelligence and Big Data require a different kind of skillset. Does India have enough skilled people to fill these roles?
Jammu: The biggest challenge we’re facing today is skillsets. We are an industrial company. We need people who understand both the engineering side of things and the computer side of things, and that is a very difficult skillset to find. So we try to hire engineers who do computer science and co-locate them with core engineers so that the domain knowledge transfer can seamlessly go back and forth.
What about the lack of standards and interoperability?
Kolla: The current solutions being developed support every protocol out there so that two-three years hence when something is standardised, the existing solution will be accepted. This means that current solutions will be processor heavy, power heavy, and memory heavy, but that’s the risk we have to take.
Editor's Picks »
- What ABB India’s performance in June quarter says about capex growth
- Bajaj Finance does well in Q1 even as competition hots up
- Kotak Mahindra Bank: The perils of being priced to perfection
- Higher cane price crushes hopes of sugar mills
- Market optimism before 2019 general election: History may not repeat itself