This work investigates the theoretical boundaries of creating publicly-detectable schemes to enable the provenance of watermarked imagery. Metadata-based approaches like C2PA provide unforgeability and public-detectability. ML techniques offer robust retrieval and watermarking. However, no existing scheme combines robustness, unforgeability, and public-detectability. In this work, we formally define such a scheme and establish its existence. Although theoretically possible, we find that at present, it is intractable to build certain components of our scheme without a leap in deep learning capabilities. We analyze these limitations and propose research directions that need to be addressed before we can practically realize robust and publicly-verifiable provenance.

The ability to properly benchmark model performance in the face of spurious correlations is important to both build better predictors and increase confidence that models are operating as intended. We demonstrate that characterizing (as opposed to simply quantifying) model mistakes across subgroups is pivotal to properly reflect model biases, which are ignored by standard metrics such as worst-group accuracy or accuracy gap. Inspired by the hypothesis testing framework, we introduce SkewSize, a principled and flexible metric that captures bias from mistakes in a model's predictions. It can be used in multi-class settings or generalised to the open vocabulary setting of generative models. SkewSize is an aggregation of the effect size of the interaction between two categorical variables: the spurious variable representing the bias attribute the model's prediction. We demonstrate the utility of SkewSize in multiple settings including: standard vision models trained on synthetic data, vision models trained on ImageNet, and large scale vision-and-language models from the BLIP-2 family. In each case, the proposed SkewSize is able to highlight biases not captured by other metrics, while also providing insights on the impact of recently proposed techniques, such as instruction tuning.

Adversarial training is widely used to make classifiers robust to a specific threat or adversary, such as lp- norm bounded perturbations of a given p-norm. However, existing methods for training classifiers robust to multiple threats require knowledge of all attacks during training and remain vulnerable to unseen distribution shifts. In this work, we describe how to obtain adversarially-robust model soups (i.e., linear combinations of parameters) that smoothly trade-off robustness to different lp-norm bounded adversaries. We demonstrate that such soups allow us to control the type and level of robustness, and can achieve robustness to all threats without jointly training on all of them. In some cases, the resulting model soups are more robust to a given lp-norm adversary than the constituent model specialized against that same adversary. Finally, we show that adversarially-robust model soups can be a viable tool to adapt to distribution shifts from a few examples.

Automated content filtering and moderation is an important tool that allows online platforms to build striving user communities that facilitate cooperation and prevent abuse. Unfortunately, resourceful actors try to bypass automated filters in a bid to post content that violate platform policies and codes of conduct. To reach this goal, these malicious actors may obfuscate policy violating images (e.g., overlay harmful images by carefully selected benign images or visual patterns) to prevent machine learning models from reaching the correct decision. In this paper, we invite researchers to tackle this specific issue and present a new image benchmark. This benchmark, based on ImageNet, simulates the type of obfuscations created by malicious actors. It goes beyond Image-Net-C and ImageNet-C-bar by proposing general, drastic, adversarial modifications that preserve the original content intent. It aims to tackle a more common adversarial threat than the one considered by lp-norm bounded adversaries. We evaluate 33 pretrained models on the benchmark and train models with different augmentations, architectures and training methods on subsets of the obfuscations to measure generalization. Our hope is that this benchmark will encourage researchers to test their models and methods and try to find new approaches that are more robust to these obfuscations.

Adaptive defenses, which optimize at test time, promise to improve adversarial robustness. We categorize such adaptive test-time defenses, explain their potential benefits and drawbacks, and evaluate a representative variety of the latest adaptive defenses for image classification. Unfortunately, none significantly improve upon static defenses when subjected to our careful case study evaluation. Some even weaken the underlying static model while simultaneously increasing inference computation. While these results are disappointing, we still believe that adaptive test-time defenses are a promising avenue of research and, as such, we provide recommendations for their thorough evaluation. We extend the checklist of Carlini et al. (2019) by providing concrete steps specific to adaptive defenses.

We introduce the Lossy Implicit Network Activation Coding (LINAC) defence, an input transformation which successfully hinders several common adversarial attacks on CIFAR-10 classifiers for perturbations up to 8/255 in Linf norm and 0.5 in L2 norm. Implicit neural representations are used to approximately encode pixel colour intensities in 2D images such that classifiers trained on transformed data appear to have robustness to small perturbations without adversarial training or large drops in performance. The seed of the random number generator used to initialise and train the implicit neural representation turns out to be necessary information for stronger generic attacks, suggesting its role as a private key. We devise a Parametric Bypass Approximation (PBA) attack strategy for key-based defences, which successfully invalidates an existing method in this category. Interestingly, our LINAC defence also hinders some transfer and adaptive attacks, including our novel PBA strategy. Our results emphasise the importance of a broad range of customised attacks despite apparent robustness according to standard evaluations.

Most real world applications require dealing with stochasticity like sensor noise or predictive uncertainty, where formal specifications of desired behavior are inherently probabilistic. Despite the promise of formal verification in ensuring the reliability of neural networks, progress in the direction of probabilistic specifications has been limited. In this direction, we first introduce a general formulation of probabilistic specifications for neural networks, which captures both probabilistic networks (e.g., Bayesian neural networks, MC-Dropout networks) and uncertain inputs (distributions over inputs arising from sensor noise or other perturbations). We then propose a general technique to verify such specifications by generalizing the notion of Lagrangian duality, replacing standard Lagrangian multipliers with "functional multipliers" that can be arbitrary functions of the activations at a given layer. We show that an optimal choice of functional multipliers leads to exact verification (i.e., sound and complete verification), and for specific forms of multipliers, we develop tractable practical verification algorithms. We empirically validate our algorithms by applying them to Bayesian Neural Networks (BNNs) and MC Dropout Networks, and certifying properties such as adversarial robustness and robust detection of out-of-distribution (OOD) data. On these tasks we are able to provide significantly stronger guarantees when compared to prior work -- for instance, for a VGG-64 MC-Dropout CNN trained on CIFAR-10 in a verification-agnostic manner, we improve the certified AUC (a verified lower bound on the true AUC) for robust OOD detection (on CIFAR-100) from $0 \% \rightarrow 29\%$. Similarly, for a BNN trained on MNIST, we improve on the $\ell_\infty$ robust accuracy from $60.2 \% \rightarrow 74.6\%$. Further, on a novel specification -- distributionally robust OOD detection -- we improve on the certified AUC from $5\% \rightarrow 23\%$.

Recent work argues that robust training requires substantially larger datasets than those required for standard classification. On CIFAR-10 and CIFAR-100, this translates into a sizable robust-accuracy gap between models trained solely on data from the original training set and those trained with additional data extracted from the "80 Million Tiny Images" dataset (TI-80M). In this paper, we explore how generative models trained solely on the original training set can be leveraged to artificially increase the size of the original training set and improve adversarial robustness to $\ell_p$ norm-bounded perturbations. We identify the sufficient conditions under which incorporating additional generated data can improve robustness, and demonstrate that it is possible to significantly reduce the robust-accuracy gap to models trained with additional real data. Surprisingly, we even show that even the addition of non-realistic random data (generated by Gaussian sampling) can improve robustness. We evaluate our approach on CIFAR-10, CIFAR-100, SVHN and TinyImageNet against $\ell_\infty$ and $\ell_2$ norm-bounded perturbations of size $\epsilon = 8/255$ and $\epsilon = 128/255$, respectively. We show large absolute improvements in robust accuracy compared to previous state-of-the-art methods. Against $\ell_\infty$ norm-bounded perturbations of size $\epsilon = 8/255$, our models achieve 66.10% and 33.49% robust accuracy on CIFAR-10 and CIFAR-100, respectively (improving upon the state-of-the-art by +8.96% and +3.29%). Against $\ell_2$ norm-bounded perturbations of size $\epsilon = 128/255$, our model achieves 78.31% on CIFAR-10 (+3.81%). These results beat most prior works that use external data.

Adversarial training suffers from robust overfitting, a phenomenon where the robust test accuracy starts to decrease during training. In this paper, we focus on reducing robust overfitting by using common data augmentation schemes. We demonstrate that, contrary to previous findings, when combined with model weight averaging, data augmentation can significantly boost robust accuracy. Furthermore, we compare various augmentations techniques and observe that spatial composition techniques work the best for adversarial training. Finally, we evaluate our approach on CIFAR-10 against $\ell_\infty$ and $\ell_2$ norm-bounded perturbations of size $\epsilon = 8/255$ and $\epsilon = 128/255$, respectively. We show large absolute improvements of +2.93% and +2.16% in robust accuracy compared to previous state-of-the-art methods. In particular, against $\ell_\infty$ norm-bounded perturbations of size $\epsilon = 8/255$, our model reaches 60.07% robust accuracy without using any external data. We also achieve a significant performance boost with this approach while using other architectures and datasets such as CIFAR-100, SVHN and TinyImageNet.

Recent research has made the surprising finding that state-of-the-art deep learning models sometimes fail to generalize to small variations of the input. Adversarial training has been shown to be an effective approach to overcome this problem. However, its application has been limited to enforcing invariance to analytically defined transformations like lp-norm bounded perturbations. Such perturbations do not necessarily cover plausible real-world variations that preserve the semantics of the input (such as a change in lighting conditions). In this paper, we propose a novel approach to express and formalize robustness to these kinds of real-world transformations of the input. The two key ideas underlying our formulation are (1) leveraging disentangled representations of the input to define different factors of variations, and (2) generating new input images by adversarially composing the representations of different images. We use a StyleGAN model to demonstrate the efficacy of this framework. Specifically, we leverage the disentangled latent representations computed by a StyleGAN model to generate perturbations of an image that are similar to real-world variations (like adding make-up, or changing the skin-tone of a person) and train models to be invariant to these perturbations. Extensive experiments show that our method improves generalization and reduces the effect of spurious correlations (reducing the error rate of a "smile" detector by 21% for example).

In this paper we propose to augment a modern neuralnetwork architecture with an attention model inspired by human perception. Specifically, we adversarially train and analyze a neural model incorporating a human inspired, visual attention component that is guided by a recurrent top-down sequential process. Our experimental evaluation uncovers several notable findings about the robustness and behavior of this new model. First, introducing attention to the model significantly improves adversarial robustness resulting in state-of-the-art ImageNet accuracies under a wide range of random targeted attack strengths. Second, we show that by varying the number of attention steps (glances/fixations) for which the model is unrolled, we are able to make its defense capabilities stronger, even in light of stronger attacks - resulting in a “computational race” between the attacker and the defender. Finally, we show that some of the adversarial examples generated by attacking our model are quite different from conventional adversarial examples - they contain global, salient and spatially coherent structures coming from the target class that would be recognizable even to a human, and work by distracting the attention of the model away from the main object in the original image.

Does a Variational AutoEncoder (VAE) consistently encode typical samples generated from its decoder? This paper shows that the perhaps surprising answer to this question is `No'; a (nominally trained) VAE does not necessarily amortize inference for typical samples that it is capable of generating. We study the implications of this behaviour on the learned representations and also the consequences of fixing it by introducing a notion of self consistency. Our approach hinges on an alternative construction of the variational approximation distribution to the true posterior of an extended VAE model with a Markov chain alternating between the encoder and the decoder. The method can be used to train a VAE model from scratch or given an already trained VAE, it can be run as a post processing step in an entirely self supervised way without access to the original training data. Our experimental analysis reveals that encoders trained with our self-consistency approach lead to representations that are robust (insensitive) to perturbations in the input introduced by adversarial attacks. We provide experimental results on the ColorMnist and CelebA benchmark datasets that quantify the properties of the learned representations and compare the approach with a baseline that is specifically trained for the desired property.

Po-Sen Huang, Robert Stanforth, Johannes Welbl, Chris Dyer, Dani Yogatama, Sven Gowal, Krishnamurthy Dvijotham, Pushmeet Kohli. Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP). 2019.

Recent work has shown that it is possible to train deep neural networks that are provably robust to norm-bounded adversarial perturbations. Most of these methods are based on minimizing an upper bound on the worst-case loss over all possible adversarial perturbations. While these techniques show promise, they often result in difficult optimization procedures that remain hard to scale to larger networks. Through a comprehensive analysis, we show how a simple bounding technique, interval bound propagation (IBP), can be exploited to train large provably robust neural networks that beat the state-of-the-art in verified accuracy. While the upper bound computed by IBP can be quite weak for general networks, we demonstrate that an appropriate loss and clever hyper-parameter schedule allow the network to adapt such that the IBP bound is tight. This results in a fast and stable learning algorithm that outperforms more sophisticated methods and achieves state-of-the-art results on MNIST, CIFAR-10 and SVHN. It also allows us to train the largest model to be verified beyond vacuous bounds on a downscaled version of IMAGENET.

Predicting delayed outcomes is an important problem in recommender systems (e.g., if customers will finish reading an ebook). We formalize the problem as an adversarial, delayed online learning problem and consider how a proxy for the delayed outcome (e.g., if customers read a third of the book in 24 hours) can help minimize regret, even though the proxy is not available when making a prediction. Motivated by our regret analysis, we propose two neural network architectures: Factored Forecaster (FF) which is ideal if the proxy is informative of the outcome in hindsight, and Residual Factored Forecaster (RFF) that is robust to a non-informative proxy. Experiments on two real-world datasets for predicting human behavior show that RFF outperforms both FF and a direct forecaster that does not make use of the proxy. Our results suggest that exploiting proxies by factorization is a promising way to mitigate the impact of long delays in human-behavior prediction tasks.

This paper addressed the problem of formally verifying desirable properties of neural networks, i.e., obtaining provable guarantees that neural networks satisfy specifications relating their inputs and outputs (e.g., robustness to bounded norm adversarial perturbations). Most previous work on this topic was limited in its applicability by the size of the network, network architecture and the complexity of properties to be verified. In contrast, our framework applies to a general class of activation functions and specifications. We formulate verification as an optimization problem (seeking to find the largest violation of the specification) and solve a Lagrangian relaxation of the optimization problem to obtain an upper bound on the worst case violation of the specification being verified. Our approach is anytime, i.e., it can be stopped at any time and a valid bound on the maximum violation can be obtained. Finally, we highlight how this approach can be used to train models that are amenable to verification.

Adversarial training is an effective methodology for training deep neural networks that are robust against adversarial, norm-bounded perturbations. However, the computational cost of adversarial training grows prohibitively as the size of the model and number of input dimensions increase. Further, training against less expensive and therefore weaker adversaries produces models that are robust against weak attacks but break down under attacks that are stronger. This is often attributed to the phenomenon of gradient obfuscation; such models have a highly non-linear loss surface in the vicinity of training examples, making it hard for gradient-based attacks to succeed even though adversarial examples still exist. In this work, we introduce a novel regularizer that encourages the loss to behave linearly in the vicinity of the training data, thereby penalizing gradient obfuscation while encouraging robustness. We show via extensive experiments on CIFAR-10 and ImageNet, that models trained with our regularizer avoid gradient obfuscation and can be trained significantly faster than adversarial training. Using this regularizer, we exceed current state of the art and achieve 47% adversarial accuracy for ImageNet with L-infinity norm adversarial perturbations of radius 4/255 under an untargeted, strong, white-box attack. Additionally, we match state of the art results for CIFAR-10 at 8/255.

Since the 1960s, consensus problems have puzzled the minds of many researchers in fields ranging from computer science to information aggregation. In this work, although specifically addressing the rendezvous problem for a team of mobile robots, we develop a methodology that can also be applied to other consensus problems, where optimality is important and where non-holonomicity characterizes the system at hand. In particular, we consider a group of differential-wheeled robots endowed with noisy relative positioning capabilities. We develop a distributed, real-time optimization method based on a receding horizon controller that minimizes a user-defined cost whilst guaranteeing the rendezvous. Finally, we perform experiments on real robots to confirm the validity of our approach.

Collision Warning Systems (CWS) are safety systems designed to warn the driver about an imminent collision. A CWS monitors the dynamic state of the traffic in realtime by processing information from various proprioceptive and exteroceptive sensors. It assesses the potential threat level and decides whether a warning should be issued to the driver through auditory and/or visual signals. Several measures have already been defined for threat assessment and various CWS have been proposed in literature. In this paper, we will focus on two time-based measures that assess both front and rear collision threats. In particular, a new threat metric, the time-to-last-second-acceleration (Tlsa), for lead vehicles in rear-end collision is proposed and compared with its counterpart, the time-to-last-second-braking (Tlsb) [18]. The Tlsais a novel time-based approach that focuses on the lead vehicle (as opposed to the following vehicle). It inherits the properties of the Tlsband, as such, is coherent with the human judgement of urgency and severity of threats. It directly quantifies the threat level of the current dynamic situation before a required evasive action (i.e. maximum acceleration) needs to be applied. Furthermore, different warning thresholds are proposed by considering the average driver reaction time. Its effect on decreasing the severity of a rear-end collision is studied and its reliability is tested using a well-established physics-based robotics simulator, namely Webots [13].

In this paper, we state, using thorough mathematical analysis, sufficient conditions to perform a rendezvous maneuver with a group of differential-wheeled robots endowed with an on-board, noisy, local positioning system. In particular, we extend the existing framework of noise-free, graph-based distributed control with a layer of Bayesian reasoning allowing to solve the rendezvous problem more efficiently in presence of uncertainties and in a probabilistically sound way. Finally we perform extensive experiments with a team of four real robots, and simulation with their corresponding simulated counterpart, to confirm the benefits of our Bayesian approach.

Multi-robot systems can solve complex tasks that require the coordination of the team-member positions with respect to each other. While the development of ad-hoc relative positioning platforms embedding cheap off-the-shelf components is a practical choice, it leads not only to differences between the platforms themselves, but also to a high sensitivity to external factors. In this paper, we present a novel lightweight online calibration method composed of two phases, capable of running on miniature robots with limited computational capabilities. Furthermore, by exploiting a Gaussian process regression in its second phase, the proposed calibration approach is able to capture deviations from an assumed underlying physical model. We compare the performance of our approach with the theoretical Cramér-Rao lower bound and test its efficiency on real robots equipped with range and bearing modules.

Using graph theory, this paper investigates how a group of robots, endowed with local positioning (range and bearing from other robots), can be engaged in a leader-following mission whilst keeping a predefined configuration. The possibility to locally change the behaviors of the follower team to accomodate both tasks is explored. In particular, a methodology to automatically adjust the parameters of the inter-robot interactions and a nonlinear PI controller are explained and implemented. Our approach is supported by a mathematical analysis as well as real robot experiments.

Using graph theory, this paper investigates how a group of vehicles, endowed with local positioning capabilities (range and bearing to other vehicles), can keep a predefined formation. We propose a longitudinal and lateral controller that stabilizes a system of several vehicles as well as a collision avoidance mechanism. The stability of our approach is supported by a mathematical analysis as well as realistic simulations.