From da1fbb0442d24b70583ed7412a06a75b68a0abdf Mon Sep 17 00:00:00 2001 From: Jeroen Overschie Date: Sat, 9 Jul 2022 19:29:55 +0200 Subject: [PATCH] Update README.md --- README.md | 16 ++++++++++++++-- 1 file changed, 14 insertions(+), 2 deletions(-) diff --git a/README.md b/README.md index 59aa8d9..c886f7c 100644 --- a/README.md +++ b/README.md @@ -2,13 +2,25 @@ A demonstration of implementing backdoors in Deep Neural Networks. Both a normal and a latent backdoor were implemented. -## Report +## Report & demo [![demo-example](demo-example.png)](https://dunnkers.com/neural-network-backdoors/) For a detailed report, see the [interactive documentation](https://dunnkers.com/neural-network-backdoors/). -The page is a React.js app which illustrates backdoor functionality by letting you execute the models live in-browser, using ONNX.js. Give it a try, it's pretty cool. +The page is a React.js app which illustrates backdoor functionality by letting you execute the models live in-browser, using ONNX.js. Give it a try, it's pretty cool โœจ. + +## Presentation + +### ๐Ÿงช Demo: implementing backdoors +See the [Google Slides](https://docs.google.com/presentation/d/e/2PACX-1vT8bytRkxBwlFUcUt3NDViA28TIJbBZHGxqBH3ylR8pis4GtilwFiuVp2ECM_ns4gk2rRXzlDvCVgAc/pub?start=false&loop=false&delayms=3000): + +Screenshot 2022-07-09 at 19 20 35 + +### ๐Ÿ—’ Theoretical: how *latent* backdoors work +See the [Google Slides](https://docs.google.com/presentation/d/e/2PACX-1vTjnR0zfcOWyMr6j6hWgc-rIyjAUF4YfxQzmA5NRSuAgr2XOS5l8VsHfHVhX2DRleeN1u77g3e5tvYD/pub?start=false&loop=false&delayms=3000): + +Screenshot 2022-07-09 at 19 23 17 ## Running the project source The project is divided in 3 parts: implementations for 2 backdoors and the demonstration website, which is a React.js app.