https://github.com/scaleoutsystems/fedn-client-fraud-keras
FEDn project, credit card fraud
https://github.com/scaleoutsystems/fedn-client-fraud-keras
Last synced: 12 months ago
JSON representation
FEDn project, credit card fraud
- Host: GitHub
- URL: https://github.com/scaleoutsystems/fedn-client-fraud-keras
- Owner: scaleoutsystems
- Created: 2021-05-17T11:15:04.000Z (about 5 years ago)
- Default Branch: master
- Last Pushed: 2021-09-14T14:48:41.000Z (over 4 years ago)
- Last Synced: 2025-05-31T02:10:25.068Z (about 1 year ago)
- Language: Jupyter Notebook
- Homepage:
- Size: 286 KB
- Stars: 1
- Watchers: 2
- Forks: 1
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# FEDn-client-fraud-keras
This repository contains a FEDn client implementation for federated training of a Keras autoencoder model for
credit card fraud detection using the public dataset https://www.kaggle.com/mlg-ulb/creditcardfraud.
In this example the credit card transaction dataset is divided (IID) into a configurable number of clients.
In each round update, the client trains the autoencoder model locally for one epoch using a batch size of 32 (Adam)
optimizer.
To download and prepare the partitioned dataset:
$ python create_data_partitions.py
This downloads the full dataset into the file 'data.csv', and creates random partitions in data/clients/.
## Start a client
Clone this repository and then download the client.yaml config file from the FEDn UI (Network view) and copy it into the main repostitory folder.
To build the client environment:
$ docker build . -t fraud-client:latest
To start a client (using Docker):
$ docker run -it -v $(pwd)/data/clients/0:/app/data -v $(pwd)/client.yaml:/app/client.yaml fraud-client:latest fedn run client -in client.yaml