From 4811cbca85c5b7620341aa94416e4d70f12b770d Mon Sep 17 00:00:00 2001 From: Yi Zhou Date: Fri, 28 Aug 2020 10:59:47 -0700 Subject: [PATCH] update krum readme with parameter details. --- examples/krum/README.md | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/examples/krum/README.md b/examples/krum/README.md index 970dc9a..e10e201 100644 --- a/examples/krum/README.md +++ b/examples/krum/README.md @@ -3,10 +3,12 @@ **Krum proposed in: [Machine Learning with Adversaries: Byzantine Tolerant Gradient Descent](https://papers.nips.cc/paper/6617-machine-learning-with-adversaries-byzantine-tolerant-gradient-descent.pdf)** -This example explains how to run the KRUM fusion algorithm on CNNs implemented with Keras training +This example explains how to run the Krum fusion algorithm to train CNNs implemented with Keras training on [MNIST](http://yann.lecun.com/exdb/mnist/) data. Data in this example preprocessed by scaling down to range from `[0, 255]` to `[0, 1]`. No other preprocessing is performed. +**Note that** the parameter `byzantine_threshold` is set to `1` by default. According to Krum's assumption, the number of parties in the FL system should be at least `2 * byzantine_threshold + 3`. Therefore, to run this example, please create at least `5` parties. + - Split data by running: ```