Speaker "Sharmistha Chatterjee" Details Back



Scaling privacy embedded  distributed ML systems


Machine learning has played an increasingly important role in big data due to its capability of efficiently extracting meaningful information and adding valuable knowledge to large diverse systems. Data from multiple organizations may exhibit different privacy policies and requirements, related to sharing data publicly while processing, assimilating, and training the data are prerequisites for any scalable architecture. As data distribution and sharing gained prominence, ML research felt the need of preserving privacy in ML models so as to prevent leakage of sensitive information outside. In this context, we organize our talk on several mechanisms to build ML models without violating privacy concerns.
We focus our talk related to privacy-preserving ML, encryption mechanisms, membership inference attacks, differential privacy for large-scale distributed systems with a practical example of its usage in the IOT industry. We also demonstrate some hands-on samples of embedding privacy using deep learning frameworks with Keras and Tensorflow libraries. 


Sharmistha Chatterjee is a Data Science Evangelist with vast professional experience in the field of Machine Learning and Cloud applications. She is currently working as a Senior Manager of Data Sciences at Publicis Sapient where she leads the digital transformation of clients across industry verticals ranging from Media, Travel and Hospitality, Advertising, IOT, and Telecom. She has proven experience in doing AI research and implementing scalable AI solutions for enterprises. In this process, she tries to bridge the gap between theory and practice. She is an active blogger (hackernoon, datasciencecentral, and an international speaker at various tech conferences. She is also a Google Developer Expert in the field of Machine Learning and one of the Hackernoon Tech award winners for 2020. She enjoys mentoring and, teaching others helping them to learn and grow.