Podcast: Anomaly Detection on Live Data

Podcast: Anomaly Detection on Live Data

by Randy Ksar - Developer Community Manager

On episode 3 of The Live Data Podcast, Sarit and I were joined by Arun Kejariwal, statistical learning principal and Francois Orsini, CTO of Satori. They are speaking at Strata Data NYC on September 28 at a session titled “Anomaly detection on live data”. Both gentlemen have lots of experience with Arun working previously at Twitter and Netflix and Francois working at Sun and Sybase. Fun fact, Francois got a job at Satori while his daughter was playing one of MZ’s mobile games.

A couple of key highlights from our conversation:

  • Live data processing complements big data. Big data aims at augmenting BI by extracting the signal out of all the data that you get. Live data processing aims at data acquisition, data processing and reaction in real time.

  • How has Anomaly Detection (AD) changed as we move from big data to live data processing?  AD is traditionally applied offline on static data, for example an anomalous product on the assembly line. Nowadays, when the data velocity is much larger, live AD methods are used on live video streams to detect ,for example, a biker on a freeway.

And here is the timeline of the podcast for your listening pleasure:

  • 3:25 - 5:52 Francois: What is your background & how did you get to Satori?

  • 5:53 -  7:44 Arun: What do you do at MZ and how did you get to where you are today?

  • 7:45 -  8:46  What are you talking about at the Strata Data conference?

  • 8:49 - 11:57  What is streaming live data and why is live data important for cities?

  • 11:58 - 14:47 Traditional vs live data anomaly detection

  • 20:32 - 24:48 Where do you see streaming live data 5 years from now?

Listen to the podcast below and hope to see some of you at Strata Data NYC 2017.


Make sure to spread the word about our session by liking or RT'ing the below. Thanks!