{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/deep-q-learning-for-fooling-neural-networks","title":"Deep Q learning for fooling neural networks","arxiv_id":"1811.05521","date":"2018-11-13","proceeding":null,"authors":["Mandar Kulkarni"],"abstract":"Deep learning models are vulnerable to external attacks. In this paper, we\npropose a Reinforcement Learning (RL) based approach to generate adversarial\nexamples for the pre-trained (target) models. We assume a semi black-box\nsetting where the only access an adversary has to the target model is the class\nprobabilities obtained for the input queries. We train a Deep Q Network (DQN)\nagent which, with experience, learns to attack only a small portion of image\npixels to generate non-targeted adversarial images. Initially, an agent\nexplores an environment by sequentially modifying random sets of image pixels\nand observes its effect on the class probabilities. At the end of an episode,\nit receives a positive (negative) reward if it succeeds (fails) to alter the\nlabel of the image. Experimental results with MNIST, CIFAR-10 and Imagenet\ndatasets demonstrate that our RL framework is able to learn an effective attack\npolicy.","url_abs":"http://arxiv.org/abs/1811.05521v1","url_pdf":"http://arxiv.org/pdf/1811.05521v1.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"deep-q-learning-for-fooling-neural-networks","repo_url":"https://github.com/mandareln/deep-q-learning-adversarial","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"tf","reach":null}],"tasks":[{"task_slug":"q-learning","task_name":"Q-Learning"},{"task_slug":"reinforcement-learning","task_name":"Reinforcement Learning"},{"task_slug":"reinforcement-learning-1","task_name":"Reinforcement Learning (RL)"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":null,"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}