Dartmouth Events

Secure and Privacy-preserving Data-driven Systems

Shagufta will present how to effectively detect crypto-ransomware and insider cyber attacks at an early stage by using machine learning techniques as one of the key components.

3/6/2020
3:30 pm – 5:00 pm
Kemeny Hall 007
Intended Audience(s): Public
Categories: Lectures & Seminars

Abstract: Given the increasing volume of sensitive data stored on systems that are connected to the internet, it is likely that cyber threats such as insider and ransomware will continue to be lucrative ammunitions for cybercriminals and cause billions of dollars of damage. While machine learning, more specifically, anomaly detection techniques can help us quickly detect and resolve such unexpected intrusions, critical applications such as health monitoring devices cannot directly leverage third-party anomaly detection services due to the sensitive nature of the data. With the growing popularity of machine learning as-a-service APIs, it is also equally important to understand if these APIs are introducing new attack vectors against the privacy of the data on which the models were trained. In this talk, I will first present how to effectively detect crypto-ransomware and insider cyber attacks at an early stage by using machine learning techniques as one of the key components. I will then discuss some of the applications that deal with sensitive data and how to design a privacy-preserving anomaly detection framework for edge computing that enables real-time detection of anomalous data in Internet-of-Things and cyber-physical systems. Finally, I will conclude with a discussion on challenges in building practicable data-driven systems that take into account both data security and privacy while also keeping the intended functionality of the system unimpaired.

Bio: Shagufta Mehnaz is a PhD candidate in Computer Science at Purdue University. She is broadly interested in the areas of security, privacy, and machine learning. Her research efforts leverage machine learning techniques to protect data from different types of adversaries as well as enhance the privacy and security of machine learning techniques and models themselves. Her papers have received CODASPY 2017 best paper award and EWSN 2017 best paper award nomination. She has been selected as one of the 100 Computer Science Young Researchers from all over the world to attend the Heidelberg Laureate Forum (HLF) in 2018 and also is a recipient of the Schlumberger Foundation Faculty For The Future (FFTF) fellowship.

 

For more information, contact:
Susan Cable

Events are free and open to the public unless otherwise noted.