Comprehending long visual documents, where information is distributed across extensive pages of text and visual elements, is a critical but challenging task for modern Vision-Language Models (VLMs).Existing approaches falter on a fundamental challenge: evidence localization.They struggle to retrieve relevant pages and overlook fine-grained details within visual elements, leading to limited performance and model hallucination.To address this, we propose DOCLENS, a toolaugmented multi-agent framework that effectively "zooms in" on evidence like a lens.It first navigates from the full document to specific visual elements on relevant pages, then employs a sampling-adjudication mechanism to generate a single, reliable answer.Paired with Gemini-2.5-Pro,DOCLENS achieves stateof-the-art performance on MMLongBench-Doc and FinRAGBench-V, surpassing even human experts.The framework's superiority is particularly evident on vision-centric and unanswerable queries, demonstrating the power of its enhanced localization capabilities.

Retrieval-augmented generation (RAG), while effective in integrating external knowledge to enhance large language models (LLMs), can be undermined by imperfect retrieval, which may introduce irrelevant, misleading, or even malicious information.Despite its importance, previous studies have rarely explored the behavior of RAG with errors from imperfect retrieval, and how potential conflicts arise between the LLMs' internal knowledge and external sources.We show that imperfect retrieval augmentation might be inevitable and quite harmful, through controlled analysis under realistic conditions.Knowledge conflicts between LLM-internal and external knowledge from retrieval is a bottleneck to overcome in the post-retrieval stage of RAG.To render LLMs resilient to imperfect retrieval, we propose ASTUTE RAG, a novel RAG approach that adaptively elicits essential information from LLMs' internal knowledge, iteratively consolidates internal and external knowledge with source-awareness, and finalizes the answer according to information reliability.Our experiments with Gemini and Claude demonstrate that ASTUTE RAG significantly outperforms previous robustness-enhanced RAG methods.Notably, ASTUTE RAG is the only approach that matches or exceeds the performance of LLMs without RAG under worst-case scenarios.ASTUTE RAG effectively resolves knowledge conflicts, improving the reliability and trustworthiness of RAG systems.45.4% LLM correct RAG correct LLM incorrect RAG incorrect Both sides are wrong.It is hard to improve, but combining internal and external knowledge may help.Previous work leverages RAG to address LLMs' knowledge gap.Zonia receives from Reuben a letter in the play.Zonia receives from Reuben a kiss in the play

Agents based on large language models (LLMs) for machine learning engineering (MLE) can automatically implement ML models via code generation. However, existing approaches to build such agents often rely heavily on inherent LLM knowledge and employ coarse exploration strategies that modify the entire code structure at once. This limits their ability to select effective task-specific models and perform deep exploration within specific components, such as experimenting extensively with feature engineering options. To overcome these, we propose MLE-STAR, a novel approach to build MLE agents. MLE-STAR first leverages external knowledge by using a search engine to retrieve effective models from the web, forming an initial solution, then iteratively refines it by exploring various strategies targeting specific ML components. This exploration is guided by ablation studies analyzing the impact of individual code blocks. Furthermore, we introduce a novel ensembling method using an effective strategy suggested by MLE-STAR. Our experimental results show that MLE-STAR achieves medals in 64% of the Kaggle competitions on the MLE-bench, significantly outperforming the best alternative.

Both transduction and rejection have emerged as important techniques for defending against adversarial perturbations. A recent work by Goldwasser et. al showed that rejection combined with transduction can give *provable* guarantees (for certain problems) that cannot be achieved otherwise. Nevertheless, under recent strong adversarial attacks (GMSA), Goldwasser et al.'s work was shown to have low performance in a practical deep-learning setting. In this paper, we take a step towards realizing the promise of transduction+rejection in more realistic scenarios. Our key observation is that a novel application of a reduction technique by Tramèr, which was until now only used to demonstrate the vulnerability of certain defenses, can be used to actually construct effective defenses. Theoretically, we show that a careful application of this technique in the transductive setting can give significantly improved sample-complexity for robust generalization. Our theory guides us to design a new transductive algorithm for learning a selective model; extensive experiments using state of the art attacks (AutoAttack, GMSA) show that our approach provides significantly better robust accuracy (81.6% on CIFAR-10 and 57.9% on CIFAR-100 under $l_\infty$ with budget 8/255) than existing techniques. The implementation is available at https://github.com/nilspalumbo/transduction-rejection.

Recently, there is an emerging interest in adversarially training a classifier with a rejection option (also known as a selective classifier) for boosting adversarial robustness. While rejection can incur a cost in many applications, existing studies typically associate zero cost with rejecting perturbed inputs, which can result in the rejection of numerous slightly-perturbed inputs that could be correctly classified. In this work, we study adversarially-robust classification with rejection in the stratified rejection setting, where the rejection cost is modeled by rejection loss functions monotonically non-increasing in the perturbation magnitude. We theoretically analyze the stratified rejection setting and propose a novel defense method -- Adversarial Training with Consistent Prediction-based Rejection (CPR) -- for building a robust selective classifier. Experiments on image datasets demonstrate that the proposed method significantly outperforms existing methods under strong adaptive attacks. For instance, on CIFAR-10, CPR reduces the total robust loss (for different rejection losses) by at least 7.3% under both seen and unseen attacks.

Out-of-distribution (OOD) detection plays a crucial role in ensuring the safe deployment of deep neural network (DNN) classifiers. While a myriad of methods have focused on improving the performance of OOD detectors, a critical gap remains in interpreting their decisions. We help bridge this gap by providing explanations for OOD detectors based on learned high-level concepts. We first propose two new metrics for assessing the effectiveness of a particular set of concepts for explaining OOD detectors: 1) detection completeness, which quantifies the sufficiency of concepts for explaining an OOD-detector's decisions, and 2) concept separability, which captures the distributional separation between in-distribution and OOD data in the concept space. Based on these metrics, we propose an unsupervised framework for learning a set of concepts that satisfy the desired properties of high detection completeness and concept separability, and demonstrate its effectiveness in providing concept-based explanations for diverse off-the-shelf OOD detectors. We also show how to identify prominent concepts contributing to the detection results, and provide further reasoning about their decisions.

This paper investigates an adversary's ease of attack in generating adversarial examples for real-world scenarios. We address three key requirements for practical attacks for the real-world: 1) automatically constraining the size and shape of the attack so it can be applied with stickers, 2) transform-robustness, i.e., robustness of a attack to environmental physical variations such as viewpoint and lighting changes, and 3) supporting attacks in not only white-box, but also black-box hard-label scenarios, so that the adversary can attack proprietary models. In this work, we propose GRAPHITE, an efficient and general framework for generating attacks that satisfy the above three key requirements. GRAPHITE takes advantage of transform-robustness, a metric based on expectation over transforms (EoT), to automatically generate small masks and optimize with gradient-free optimization. GRAPHITE is also flexible as it can easily trade-off transform-robustness, perturbation size, and query count in black-box settings. On a GTSRB model in a hard-label black-box setting, we are able to find attacks on all possible 1,806 victim-target class pairs with averages of 77.8% transform-robustness, perturbation size of 16.63% of the victim images, and 126K queries per pair. For digital-only attacks where achieving transform-robustness is not a requirement, GRAPHITE is able to find successful small-patch attacks with an average of only 566 queries for 92.2% of victim-target pairs. GRAPHITE is also able to find successful attacks using perturbations that modify small areas of the input image against PatchGuard, a recently proposed defense against patch-based attacks.

Intelligent Internet of Things (IoT) systems based on deep neural networks (DNNs) have been widely deployed in the real world. However, DNNs are found to be vulnerable to adversarial examples, which raises people’s concerns about intelligent IoT systems’ reliability and security. Testing and evaluating the robustness of IoT systems become necessary and essential. Recently, various attacks and strategies have been proposed, but the efficiency problem remains unsolved properly. Existing methods are either computationally extensive or time consuming, which is not applicable in practice. In this article, we propose a novel framework, called attack-inspired generative adversarial networks (AI-GAN) to generate adversarial examples conditionally. Once trained, it can generate adversarial perturbations efficiently given input images and target classes. We apply AI-GAN on different data sets in white-box settings, black-box settings, and targeted models protected by state-of-the-art defenses. Through extensive experiments, AI-GAN achieves high attack success rates, outperforming existing methods, and reduces generation time significantly. Moreover, for the first time, AI-GAN successfully scales to complex data sets, e.g., CIFAR-100 and ImageNet, with about 90% success rates among all classes.

Deep neural networks (DNNs) are vulnerable to adversarial examples, which are crafted by adding imperceptible perturbations to inputs. Recently different attacks and strategies have been proposed, but how to generate adversarial examples perceptually realistic and more efficiently remains unsolved. This paper proposes a novel framework called Attack-Inspired GAN (AI-GAN), where a generator, a discriminator, and an attacker are trained jointly. Once trained, it can generate adversarial perturbations efficiently given input images and target classes. Through extensive experiments on several popular datasets e.g., MNIST and CFAR-10, AI-GAN achieves high attack success rates and reduces generation time significantly in various settings. Moreover, for the first time, AI-GAN successfully scales to complicated datasets e.g., CFAR-100 with around 90% success rates among all classes.

When a deep learning model is deployed in the wild, it can encounter test data drawn from distributions different from the training data distribution and suffer drop in performance. For safe deployment, it is essential to estimate the accuracy of the pre-trained model on the test data. However, the labels for the test inputs are usually not immediately available in practice, and obtaining them can be expensive. This observation leads to two challenging tasks: (1) unsupervised accuracy estimation, which aims to estimate the accuracy of a pre-trained classifier on a set of unlabeled test inputs; (2) error detection, which aims to identify mis-classified test inputs. In this paper, we propose a principled and practically effective framework that simultaneously addresses the two tasks. The proposed framework iteratively learns an ensemble of models to identify mis-classified data points and performs self-training to improve the ensemble with the identified points. Theoretical analysis demonstrates that our framework enjoys provable guarantees for both accuracy estimation and error detection under mild conditions readily satisfied by practical deep learning models. Along with the framework, we proposed and experimented with two instantiations and achieved state-of-the-art results on 59 tasks. For example, on iWildCam, one instantiation reduces the estimation error for unsupervised accuracy estimation by at least 70% and improves the F1 score for error detection by at least 4.7% compared to existing methods.

We show new connections between adversarial learning and explainability for deep neural networks (DNNs). One form of explanation of the output of a neural network model in terms of its input features, is a vector of feature-attributions, which can be generated by various techniques such as Integrated Gradients (IG), DeepSHAP, LIME, and CXPlain. Two desirable characteristics of an attribution-based explanation are: (1) \emph{sparseness}: the attributions of irrelevant or weakly relevant features should be negligible, thus resulting in \emph{concise} explanations in terms of the significant features, and (2) \emph{stability}: it should not vary significantly within a small local neighborhood of the input. Our first contribution is a theoretical exploration of how these two properties (when using IG-based attributions) are related to adversarial training, for a class of 1-layer networks (which includes logistic regression models for binary and multi-class classification); for these networks we show that (a) adversarial training using an $\ell_\infty$-bounded adversary produces models with sparse attribution vectors, and (b) natural model-training while encouraging stable explanations (via an extra term in the loss function), is equivalent to adversarial training. Our second contribution is an empirical verification of phenomenon (a), which we show, somewhat surprisingly, occurs \emph{not only in 1-layer networks, but also DNNs trained on standard image datasets}, and extends beyond IG-based attributions, to those based on DeepSHAP: adversarial training with $\linf$-bounded perturbations yields significantly sparser attribution vectors, with little degradation in performance on natural test data, compared to natural training. Moreover, the sparseness of the attribution vectors is significantly better than that achievable via $\ell_1$-regularized natural training.

Wide adoption of artificial neural networks in various domains has led to an increasing interest in defending adversarial attacks against them. Preprocessing defense methods such as pixel discretization are particularly attractive in practice due to their simplicity, low computational overhead, and applicability to various systems. It is observed that such methods work well on simple datasets like MNIST, but break on more complicated ones like ImageNet under recently proposed strong white-box attacks. To understand the conditions for success and potentials for improvement, we study the pixel discretization defense method, including more sophisticated variants that take into account the properties of the dataset being discretized. Our results again show poor resistance against the strong attacks. We analyze our results in a theoretical framework and offer strong evidence that pixel discretization is unlikely to work on all but the simplest of the datasets. Furthermore, our arguments present insights why some other preprocessing defenses may be insecure.

An emerging problem in trustworthy machine learning is to train models that produce robust interpretations for their predictions. We take a step towards solving this problem through the lens of axiomatic attribution of neural networks. Our theory is grounded in the recent work, Integrated Gradients (IG) [STY17], in axiomatically attributing a neural network’s output change to its input change. We propose training objectives in classic robust optimization models to achieve robust IG attributions. Our objectives give principled generalizations of previous objectives designed for robust predictions, and they naturally degenerate to classic soft-margin training for one-layer neural networks. We also generalize previous theory and prove that the objectives for different robust optimization models are closely related. Experiments demonstrate the effectiveness of our method, and also point to intriguing problems which hint at the need for better optimization techniques or better neural network architectures for robust attribution training.

In this paper we study leveraging confidence information induced by adversarial training to reinforce adversarial robustness of a given adversarially trained model. A natural measure of confidence is $\|F(x)\|_\infty$ (i.e. how confident $F$ is about its prediction?). We start by analyzing an adversarial training formulation proposed by Madry et al.. We demonstrate that, under a variety of instantiations, an only somewhat good solution to their objective induces confidence to be a discriminator, which can distinguish between right and wrong model predictions in a neighborhood of a point sampled from the underlying distribution. Based on this, we propose Highly Confident Near Neighbor (HCNN) a framework that combines confidence information and nearest neighbor search, to reinforce adversarial robustness of a base model. We give algorithms in this framework and perform a detailed empirical study. We report encouraging experimental results that support our analysis, and also discuss problems we observed with existing adversarial training.