This website uses cookies to ensure you get the best experience on our website. Learn more

Event Diary - Data Hack Summit 2017: The Need to Rethink Deep Learning

Posted on December 15, 2017
By Avinash Patchava
and Rajesh Pantina, Marketing Lead, Performance SolutionsMarketing Lead, Performance Solutions

InMobi regularly hosts, participates in and contributes to a series of events, meet-ups, webinars, sharing best practices with partners and thought leaders, across the globe. In this edition of the InMobi Event Diary, we are looking at the highlights and developments from “Data Hack Summit 2017,” an event organized by Analytics Vidhya in Bangalore, India.

In this edition of the InMobi event diary, we cover our presence at the Analytics Vidhya-hosted event “Data Hack Summit 2017.” Aimed at bringing together the finest data scientists across India and the globe, the Data Hack Summit covered the latest trends in analytics, machine learning (ML) and artificial intelligence (AI).As the ‘AI’ and ‘ML’ buzzwords make their way into corporate boardrooms, data scientists at the conference were busy delving into the real science behind the hype. InMobi was part of one such session on “The need to rethink Deep Learning.” The session was focused on discussing the current applications across industries, the challenges in getting impact, and the evolution of deep learning as a tool in the data scientist’s armoury.

The Need to Re-think Deep LearningAvi Patchava, VP, Machine Learning and Artificial Intelligence, India at InMobi was part of the panel discussion on “The need to rethink Deep Learning.” Moderated by Kunal Jain, Founder & CEO of Analytics Vidhya, the panel also featured an eclectic mix of other researchers and practitioners - including Dr. Kirk Borne, Principal Data Scientist at Booz Allen Hamilton, Dr. Srinivasan Parthiban, CEO of Vingyani, Vijay Gabale from Huew and Mukesh Jain, CTO of VFS Global.

Here are three interesting themes that the panel explored.

Deep Learning - Why is it Different?

The success of Deep Learning to date has been in specific use cases such as image recognition using ‘convolutional neural networks.’ A few panelists held the view that Deep Learning has been a small step forward in our journey of more sophisticated statistical learning techniques - a journey we have been on for many decades. Other panelists felt strongly that it has really pushed the abilities for solving certain types of classification problems, which would not have been possible without the unique structure and design of neural networks like the deep layers of multiple neurons. The panel, however, agreed that there remains much potential for greater application of Deep Learning in sectors such as Pharma, Manufacturing, IT services and AdTech.

New frontiers for Deep Learning Application

Dr. Srinivasan opined that Deep Learning will become more and more relevant in the pharma industry. There is an abundance of disparate information in the field of drug discovery and Deep Learning will be vital in connecting these distant pieces of medical research from around the world. Avi talked about InMobi’s application of Deep Learning to understand the effectiveness of different mobile ad creative - the piece of art that is used to convey messaging in an advertisement. These techniques leverage convolutional neural networks - a deep learning application that trains well on an image as a grid of pixels. The possibility of Deep Learning being applied to predict human behaviours at scale - such as click-through and app-download behaviours is an exciting prospect for both the industry and InMobi.

Challenges in Driving Impact through Deep Learning

Driving real business impact through Deep Learning has often had challenges such as insufficient data, privacy concerns on data usage and organizational readiness to apply Deep Learning models. Additionally, the size of big data that is needed to effectively train a Deep Learning algorithm is often underestimated. With the big data needs in Deep Learning, there are challenges and sensitivities of using more and more individual-level data in solving problems. If the data is not sufficiently big, a simpler Machine Learning algorithm is more effective. There are other self-created challenges such as undue pressure to apply the ‘latest and greatest in Deep Learning’ when it is not necessarily the most suitable choice of algorithmic approach. Lastly, Avi emphasised that getting impact from a Deep Learning model is as much about the change management process as it is about building an accurate model. The change management process involves: understanding the model use and designing it accordingly; introducing stakeholders to the basis and approach for the model; and integrating a model into the business process, or even resdesigning the business process around the model.

Commenting on the future of Deep Learning, the panelists exuded optimism around the opportunity but believed that the hype is still ahead of the achieved impact for Deep Learning in a wider set of industries.

Back to Top