Backdoor attacks inject poisoning samples during training, with the goal of forcing a machine learning model to output an attacker-chosen class when presented a specific trigger at test time. Although backdoor attacks have been demonstrated in a variety of settings and against different models, the factors affecting their effectiveness are still not well understood. In this work, we provide a unifying framework to study the process of backdoor learning under the lens of incremental learning and influence functions. We show that the effectiveness of backdoor attacks depends on: (i) the complexity of the learning algorithm, controlled by its hyperparameters; (ii) the fraction of backdoor samples injected into the training set; and (iii) the size and visibility of the backdoor trigger. These factors affect how fast a model learns to correlate the presence of the backdoor trigger with the target class. Our analysis unveils the intriguing existence of a region in the hyperparameter space in which the accuracy on clean test samples is still high while backdoor attacks are ineffective, thereby suggesting novel criteria to improve existing defenses.

Deep Neural Networks are vulnerable to adversarial examples, i.e., carefully crafted input samples that can cause models to make incorrect predictions with high confidence. To mitigate these vulnerabilities, adversarial training and detection-based defenses have been proposed to strengthen models in advance. However, most of these approaches focus on a single data modality, overlooking the relationships between visual patterns and textual descriptions of the input. In this paper, we propose a novel defense, Multi-Shield, designed to combine and complement these defenses with multi-modal information to further enhance their robustness. Multi-Shield leverages multi-modal large language models to detect adversarial examples and abstain from uncertain classifications when there is no alignment between textual and visual representations of the input. Extensive evaluations on CIFAR-10 and ImageNet datasets, using robust and non-robust image classification models, demonstrate that Multi-Shield can be easily integrated to detect and reject adversarial examples, outperforming the original defenses.

In learning problems, the noise inherent to the task at hand hinders the possibility to infer without a certain degree of uncertainty. Quantifying this uncertainty, regardless of its wide use, assumes high relevance for security-sensitive applications. Within these scenarios, it becomes fundamental to guarantee good (i.e., trustworthy) uncertainty measures, which downstream modules can securely employ to drive the final decision-making process. However, an attacker may be interested in forcing the system to produce either (i) highly uncertain outputs jeopardizing the system's availability or (ii) low uncertainty estimates, making the system accept uncertain samples that would instead require a careful inspection (e.g., human intervention). Therefore, it becomes fundamental to understand how to obtain robust uncertainty estimates against these kinds of attacks. In this work, we reveal both empirically and theoretically that defending against adversarial examples, i.e., carefully perturbed samples that cause misclassification, additionally guarantees a more secure, trustworthy uncertainty estimate under common attack scenarios without the need for an ad-hoc defense strategy. To support our claims, we evaluate multiple adversarial-robust models from the publicly available benchmark RobustBench on the CIFAR-10 and ImageNet datasets.

However, having a large parameter space is considered one of the main suspects of the neural networks' vulnerability to adversarial examples-- input samples crafted ad-hoc to induce a desired misclassification. Relevant literature has claimed contradictory remarks in support of and against the robustness of over-parameterized networks. These contradictory findings might be due to the failure of the attack employed to evaluate the networks' robustness. Previous research has demonstrated that depending on the considered model, the algorithm employed to generate adversarial examples may not function properly, leading to overestimating the model's robustness. In this work, we empirically study the robustness of over-parameterized networks against adversarial examples. However, unlike the previous works, we also evaluate the considered attack's reliability to support the results' veracity. Our results show that over-parameterized networks are robust against adversarial attacks as opposed to their under-parameterized counterparts.

Adversarial examples are typically optimized with gradient-based attacks. While novel attacks are continuously proposed, each is shown to outperform its predecessors using different experimental setups, hyperparameter settings, and number of forward and backward calls to the target models. This provides overly-optimistic and even biased evaluations that may unfairly favor one particular attack over the others. In this work, we aim to overcome these limitations by proposing AttackBench, i.e., the first evaluation framework that enables a fair comparison among different attacks. To this end, we first propose a categorization of gradient-based attacks, identifying their main components and differences. We then introduce our framework, which evaluates their effectiveness and efficiency. We measure these characteristics by (i) defining an optimality metric that quantifies how close an attack is to the optimal solution, and (ii) limiting the number of forward and backward queries to the model, such that all attacks are compared within a given maximum query budget. Our extensive experimental analysis compares more than 100 attack implementations with a total of over 800 different configurations against CIFAR-10 and ImageNet models, highlighting that only very few attacks outperform all the competing approaches. Within this analysis, we shed light on several implementation issues that prevent many attacks from finding better solutions or running at all. We release AttackBench as a publicly available benchmark, aiming to continuously update it to include and evaluate novel gradient-based attacks for optimizing adversarial examples.

Evaluating the adversarial robustness of deep networks to gradient-based attacks is challenging. While most attacks consider $\ell_2$- and $\ell_\infty$-norm constraints to craft input perturbations, only a few investigate sparse $\ell_1$- and $\ell_0$-norm attacks. In particular, $\ell_0$-norm attacks remain the least studied due to the inherent complexity of optimizing over a non-convex and non-differentiable constraint. However, evaluating adversarial robustness under these attacks could reveal weaknesses otherwise left untested with more conventional $\ell_2$- and $\ell_\infty$-norm attacks. In this work, we propose a novel $\ell_0$-norm attack, called $\sigma$-zero, which leverages an ad hoc differentiable approximation of the $\ell_0$ norm to facilitate gradient-based optimization, and an adaptive projection operator to dynamically adjust the trade-off between loss minimization and perturbation sparsity. Extensive evaluations using MNIST, CIFAR10, and ImageNet datasets, involving robust and non-robust models, show that $\sigma$-zero finds minimum $\ell_0$-norm adversarial examples without requiring any time-consuming hyperparameter tuning, and that it outperforms all competing sparse attacks in terms of success rate, perturbation size, and scalability.

Model stealing aims at inferring a victim model's functionality at a fraction of the original training cost. While the goal is clear, in practice the model's architecture, weight dimension, and original training data can not be determined exactly, leading to mutual uncertainty during stealing. In this work, we explicitly tackle this uncertainty by generating multiple possible networks and combining their predictions to improve the quality of the stolen model. For this, we compare five popular uncertainty quantification models in a model stealing task. Surprisingly, our results indicate that the considered models only lead to marginal improvements in terms of label agreement (i.e., fidelity) to the stolen model. To find the cause of this, we inspect the diversity of the model's prediction by looking at the prediction variance as a function of training iterations. We realize that during training, the models tend to have similar predictions, indicating that the network diversity we wanted to leverage using uncertainty quantification models is not (high) enough for improvements on the model stealing task.

Deep learning models undergo a significant increase in the number of parameters they possess, leading to the execution of a larger number of operations during inference. This expansion significantly contributes to higher energy consumption and prediction latency. In this work, we propose EAT, a gradient-based algorithm that aims to reduce energy consumption during model training. To this end, we leverage a differentiable approximation of the $\ell_0$ norm, and use it as a sparse penalty over the training loss. Through our experimental analysis conducted on three datasets and two deep neural networks, we demonstrate that our energy-aware training algorithm EAT is able to train networks with a better trade-off between classification performance and energy efficiency.

Sponge examples are test-time inputs carefully optimized to increase energy consumption and latency of neural networks when deployed on hardware accelerators. In this work, we are the first to demonstrate that sponge examples can also be injected at training time, via an attack that we call sponge poisoning. This attack allows one to increase the energy consumption and latency of machine-learning models indiscriminately on each test-time input. We present a novel formalization for sponge poisoning, overcoming the limitations related to the optimization of test-time sponge examples, and show that this attack is possible even if the attacker only controls a few model updates; for instance, if model training is outsourced to an untrusted third-party or distributed via federated learning. Our extensive experimental analysis shows that sponge poisoning can almost completely vanish the effect of hardware accelerators. We also analyze the activations of poisoned models, identifying which components are more vulnerable to this attack. Finally, we examine the feasibility of countermeasures against sponge poisoning to decrease energy consumption, showing that sanitization methods may be overly expensive for most of the users.

The success of machine learning is fueled by the increasing availability of computing power and large training datasets. The training data is used to learn new models or update existing ones, assuming that it is sufficiently representative of the data that will be encountered at test time. This assumption is challenged by the threat of poisoning, an attack that manipulates the training data to compromise the model's performance at test time. Although poisoning has been acknowledged as a relevant threat in industry applications, and a variety of different attacks and defenses have been proposed so far, a complete systematization and critical review of the field is still missing. In this survey, we provide a comprehensive systematization of poisoning attacks and defenses in machine learning, reviewing more than 100 papers published in the field in the last 15 years. We start by categorizing the current threat models and attacks, and then organize existing defenses accordingly. While we focus mostly on computer-vision applications, we argue that our systematization also encompasses state-of-the-art attacks and defenses for other data modalities. Finally, we discuss existing resources for research in poisoning, and shed light on the current limitations and open research questions in this research field.