Device Classification with DHCP Fingerprints using Machine Learning
Time 10/18/18 01:40PM-02:00PM
This talk will present the results of a project at the University of Michigan to create an API to classify devices as phone, tablet or computer based on the device’s DHCP request. The goal of this effort is to improve the quality of our WiFi delivery by better understanding how the service is used. We will cover how we built a labeled data set to train a supervised machine learning algorithm and created an API to return predictions for new data. We will discuss how we implemented this using operational data and how we built the tool with Python, scikit-learn, Flask and Docker.
Speaker Kristoffer Steinhoff University of Michigan - Ann Arbor
Primary track Advanced Networking