Let us start with something simple. We download a neural network to try out in a project. It might sort images or help with a data pipeline. That model looks like just a file we use and forget. Now imagine that file is carrying a hidden payload. It might sound like a thought experiment, but …
Today I will be discussing my understanding of the paper Universal Backdoor Attacks. The paper delves into another exciting exploit that can be leveraged against popular convolutional models such as Resnets. What is a Universal Backdoor A backdoor is an alternate entry to your house. In the field of computers and security in general, it …
So today I will be analyzing and providing a brief summary of my understanding of this paper titled Remember What You Want To Forget This paper presents one of the algorithms for machine unlearning. Machine Unlearning is an important practice when it comes to privacy and security in machine learning. Machine unlearning algorithms aim to …
Differential privacy is a technique devised for data analysis whilst maintaining the privacy of the individuals whose data is involved. It was first introduced in the year 2006, by Cynthia Dwork, Frank McSherry, Kobbi Nissim, and Adam Smith and has been one of the foundational pillars in the intersection of data science and privacy. At …
Previously we talked about Fast Sign Gradient Method( FGSM), we saw how this white box technique, cleverly exploits the gradients in a model, to perturb the input to give the wrong prediction from the model. Since, in this method, we perturb our input just once, a modified version of this attack does so repeatedly for …
In this post, we will be talking about the vulnerabilities that plague machine learning. Yes, in the realm of computer science, no field is void of vulnerabilities and loopholes and as we progress towards a very AI-based future, the security and robustness of machine learning models become an important aspect. What are Adversarial Attacks? The …