search

TalwalkarLab / leaf

Leaf: A Benchmark for Federated Settings
BSD 2-Clause "Simplified" License
852 stars 244 forks source link

federated learning anomaly detection #38

Open ParyaHM opened 3 years ago

ParyaHM commented 3 years ago

Hi guys, I want to use federated learning for anomaly detection and attack classification. Which model do you think best matches this problem? and how can I use my own data set on this models? Also my data set does not contain client IDs originally may I just use their number as ID?

tdye24 commented 3 years ago

Hi guys, I want to use federated learning for anomaly detection and attack classification. Which model do you think best matches this problem? and how can I use my own data set on this models? Also my data set does not contain client IDs originally may I just use their number as ID?

Yeah, just assign an unique ID for each client.

tdye24 commented 3 years ago

Hi guys, I want to use federated learning for anomaly detection and attack classification. Which model do you think best matches this problem? and how can I use my own data set on this models? Also my data set does not contain client IDs originally may I just use their number as ID?

Convert your own dataset to the proper format, {'client id': {'x': [], 'y': []}}, as can be checked in the data dir.