Smartphone Impostor Detection with Behavioral Data Privacy and Minimalist Hardware Support
Offered By: tinyML via YouTube
Course Description
Overview
Explore a defense-in-depth mechanism for detecting smartphone impostors using simple Deep Learning algorithms in this tinyML Research Symposium 2021 presentation. Delve into a privacy-preserving approach that utilizes Recurrent Neural Networks (RNNs) to learn legitimate user behavior without exposing sensor data outside the device. Discover how Prediction Error Distribution (PED) enhances detection accuracy and learn about the proposed minimalist hardware module, SID (Smartphone Imposter Detector), designed for on-device, real-time impostor detection. Examine experimental results showcasing SID's performance, including its low hardware cost and energy consumption compared to other RNN accelerators.
Syllabus
Introduction
Overview
Cloudbased scenario
Privacy preserving scenario
Workflow
Detection Accuracy
Sim Module
Syntax Module
Primitives
Execution Time
Memory Usage
Conclusion
Sponsors
Taught by
tinyML
Related Courses
FPGA computing systems: Partial Dynamic ReconfigurationPolitecnico di Milano via Polimi OPEN KNOWLEDGE Introduction to Amazon Elastic Inference
Pluralsight FPGA computing systems: Partial Dynamic Reconfiguration
Politecnico di Milano via Coursera Introduction to Amazon Elastic Inference (Traditional Chinese)
Amazon Web Services via AWS Skill Builder Introduction to Amazon Elastic Inference (Portuguese)
Amazon Web Services via AWS Skill Builder