IBM / Autozoom-Attack

Codes for reproducing query-efficient black-box attacks in “AutoZOOM: Autoencoder-based Zeroth Order Optimization Method for Attacking Black-box Neural Networks” ​​​​, published at AAAI 2019
https://arxiv.org/abs/1805.11770
Apache License 2.0
57 stars 22 forks source link

Bug report, the attacked classification model does not load any pre-trained weights! #13

Open machanic opened 4 years ago

machanic commented 4 years ago

I can't find any saver.restore(sess, weight_path) code in main.py or blackbox_attack.py for loading classifier weights. Actually, this code attack the untrained classification model, which outputs almost random probabilities prediction?

chunchentu commented 4 years ago

We provided the pretrained model in this repo.

Please make sure you utilize this repo correctly and provide example code to support your issue.