Smart Analytics in Smartphones
Submitted by Satnam Singh, PhD (@satnam74s) on Sunday, 12 May 2013
Analytics and Visualization
The objective of this talk is to present the pros and cons of performing data mining on Smartphones vs. Server. I am intending to discuss state of the art use cases that can employ data mining and machine learning techniques on Smartphones. I will present a case study to demonstrate my thoughts.
Current smartphones are having quad/octa -core processors along with more than ten sensors. Using these sensors large number of data is collected and processed. This data explosion opens up classical debate – How much data processing should be performed in device vs. server? In this talk, I would like to discuss my thoughts and findings on this question along with other interesting questions on performing “Smart Analytics”. Smart Analytics is a form of data mining and machine learning that can be done in Smartphones with specific use cases in mind. I will discuss my findings with an example of performing activity recognition in smartphones. Activity recognition uses smartphone’s accelerometer sensor values to detect physical activity of the user e.g. running, jogging, walking, sitting, climbing up/down. I will show the results on few datasets using machine learning techniques.
Currently I am working as a Senior Data Scientist in Samsung Research India-Bangalore. I enjoy in applying data mining and machine learning in real world applications. My detailed profile is available at LinkedIn. http://in.linkedin.com/pub/satnam-singh-phd/2/349/347