Talks

• Grouping similar messages using Topic modeling [ talk ]

  In machine learning and natural language processing, a topic model is a type of statistical model for discovering abstract "topics" that occur in a collection of documents. At Helpshift, we get a lot of customer support messages. We use topic modelling and more specifically the Latent Dirichlet Allocation (LDA), to classify similar message automatically based on messages. These grouped messages can be processed by the CS personnel making them more efficient.
  This talk will cover Helpshift's experience and the challenges in using LDA for grouping similar messages without any prior knowledge.

  Speaker

  Vinayak Hegde is VP of Engineering at Helpshift. In the past, he has worked at Inmobi, Akamai and Microsoft.

  Twitter: @vinayakh

• The Cookie as a Customer: An E-commerce Perspective [ talk ]

  In the brick and mortar days, the shopkeeper interacted with a live human customer with appearance, expressions and behavioural traits. These attributes influenced the way the shopkeeper pitched his goods. Flip to the current e-commerce world where you are interacting with a cookie which was dropped when a customer visited your website. How does an e-commerce store know the appearance, expressions and behavioural traits of a customer? Based on those characteristics, how does the sales pitch change? Where do these conversations happen?
  In this session, Rahul Kulkarni demonstrates with real-world examples how big data from cookies translates into appearance, expressions and behavioural traits of the customer, and most importantly how these are used to make a killer sales pitch for the customer.

  Speaker

  Rahul Kulkarniis the CPO of Sokrati and has been ex-Googler.

  LinkedIn

• Application of Machine Learning for Financial Markets prediction [ talk ]

  This session covers case studies which use some of the classical as well as cutting-edge machine learning algorithms. Due to ill-conditioning and noisy nature of financial data, there are some unique characteristics of this problem that we will focus on. Robustness of modelling methodology, averaging of models, identifying what is true improvement in prediction accuracy versus over-fitting become some of the serious issues people will need look into.

  Speaker

  Aniruddha Pant is the founder and CEO of AlgoAnalytics.

  LinkedIn

• Machine Learning in Online Advertising Domain [ talk ]

  In online advertising domain, there are various players which play a different role working on behalf of either publisher or advertiser, directly or indirectly. These players interact with each other in real-time to select the best advertisement optimizing their individual goals. Sreekanth Vempati will present key optimization and real-time impression allocation challenges solved using Machine Learning by different players in the advertising echo system.
  As a part of this talk, he will present broadly about PubMatic's work with Machine Learning and their applications. Specifically, Sreekanth will show some of the Machine Learning applications in online advertising domain, showcasing some of the problems that are being solved in PubMatic.

  Speaker

  Sreekanth Vempati is the Team Lead of Machine Learning & Algorithms at PubMatic.

  LinkedIn

• Text Analytics helping IT management get smarter [ talk ]

  In this talk, Nilesh Phadke will discuss Text Analytics and how it can be used for better IT management. The talk will cover an Introduction to Text Analytics, going over the basics of what is meant by Text Analytics followed by some of the key techniques involved in Text Analytics and how these techniques can be used to improve IT management solutions.

  Speaker

  Nilesh Phadke is the Lead Product Developer at BMC Incubator Lab. He is working on using machine learning techniques for solving problems in the ITSM domain.

  LinkedIn