Explore a groundbreaking machine learning solution to combat malicious calls over telephony networks in this IEEE conference talk. Discover how researchers developed a TouchPal user interface to collect call data, conducted a large-scale measurement study, and designed 29 effective features for predicting spam and scam calls. Learn about the extensive evaluation of state-of-the-art machine learning approaches, which demonstrated the ability to reduce unblocked malicious calls by up to 90% while maintaining high precision. Gain insights into the efficiency of the models and the results of an ablation analysis, revealing that just 10 out of 29 features can achieve comparable performance to using all features.
A Machine Learning Approach to Prevent Malicious Calls Over Telephony Networks
Overview
Syllabus
A Machine Learning Approach to Prevent Malicious Calls Over Telephony Networks
Taught by
IEEE Symposium on Security and Privacy
Tags
Related Courses
-
A Deep Learning Approach to Fast, Format-Agnostic Detection of Malicious Web Content
-
ELSA - Secure Aggregation for Federated Learning with Malicious Actors
-
On the Insecurity of Peer-to-Peer Decentralized Machine Learning
-
A Theory to Instruct Differentially-Private Learning via Clipping Bias Reduction
-
Deep in the Dark - Deep Learning-based Malware Traffic Detection without Expert Knowledge
-
SoK - Let the Privacy Games Begin! A Unified Treatment of Data Inference Privacy in Machine Learning
