When dealing with imbalanced classification data, reweighting the loss function is a standard procedure allowing to equilibrate between the true positive and true negative rates within the risk measure. Despite significant theoretical work in this area, existing results do not adequately address a main challenge within the imbalanced classification framework, which is the negligible size of one class in relation to the full sample size and the need to rescale the risk function by a probability tending to zero. To address this gap, we present two novel contributions in the setting where the rare class probability approaches zero: (1) a non asymptotic fast rate probability bound for constrained balanced empirical risk minimization, and (2) a consistent upper bound for balanced nearest neighbors estimates. Our findings provide a clearer understanding of the benefits of class-weighting in realistic settings, opening new avenues for further research in this field.

Prompting is now a dominant method for evaluating the linguistic knowledge of large language models (LLMs). While other methods directly read out models' probability distributions over strings, prompting requires models to access this internal information by processing linguistic input, thereby implicitly testing a new type of emergent ability: metalinguistic judgment. In this study, we compare metalinguistic prompting and direct probability measurements as ways of measuring models' linguistic knowledge. Broadly, we find that LLMs' metalinguistic judgments are inferior to quantities directly derived from representations. Furthermore, consistency gets worse as the prompt query diverges from direct measurements of next-word probabilities. Our findings suggest that negative results relying on metalinguistic prompts cannot be taken as conclusive evidence that an LLM lacks a particular linguistic generalization. Our results also highlight the value that is lost with the move to closed APIs where access to probability distributions is limited.

Estimating the generalization error (GE) of machine learning models is fundamental, with resampling methods being the most common approach. However, in non-standard settings, particularly those where observations are not independently and identically distributed, resampling using simple random data divisions may lead to biased GE estimates. This paper strives to present well-grounded guidelines for GE estimation in various such non-standard settings: clustered data, spatial data, unequal sampling probabilities, concept drift, and hierarchically structured outcomes. Our overview combines well-established methodologies with other existing methods that, to our knowledge, have not been frequently considered in these particular settings. A unifying principle among these techniques is that the test data used in each iteration of the resampling procedure should reflect the new observations to which the model will be applied, while the training data should be representative of the entire data set used to obtain the final model. Beyond providing an overview, we address literature gaps by conducting simulation studies. These studies assess the necessity of using GE-estimation methods tailored to the respective setting. Our findings corroborate the concern that standard resampling methods often yield biased GE estimates in non-standard settings, underscoring the importance of tailored GE estimation.

Identifying driving maneuvers plays an essential role on-board vehicles to monitor driving and driver states, as well as off-board to train and evaluate machine learning algorithms for automated driving for example. Maneuvers can be characterized by vehicle kinematics or data from its surroundings including other traffic participants. Extracting relevant maneuvers therefore requires analyzing time-series of (i) structured, multi-dimensional kinematic data, and (ii) unstructured, large data samples for video, radar, or LiDAR sensors. However, such data analysis requires scalable and computationally efficient approaches, especially for non-annotated data. In this paper, we are presenting a maneuver detection approach based on two variants of space-filling curves (Z-order and Hilbert) to detect maneuvers when passing roundabouts that do not use GPS data. We systematically evaluate their respective performance by including permutations of selections of kinematic signals at varying frequencies and compare them with two alternative baselines: All manually identified roundabouts, and roundabouts that are marked by geofences. We find that encoding just longitudinal and lateral accelerations sampled at 10Hz using a Hilbert space-filling curve is already successfully identifying roundabout maneuvers, which allows to avoid the use of potentially sensitive signals such as GPS locations to comply with data protection and privacy regulations like GDPR.

Creating large-scale and well-annotated datasets to train AI algorithms is crucial for automated tumor detection and localization. However, with limited resources, it is challenging to determine the best type of annotations when annotating massive amounts of unlabeled data. To address this issue, we focus on polyps in colonoscopy videos and pancreatic tumors in abdominal CT scans; both applications require significant effort and time for pixel-wise annotation due to the high dimensional nature of the data, involving either temporary or spatial dimensions. In this paper, we develop a new annotation strategy, termed Drag&Drop, which simplifies the annotation process to drag and drop. This annotation strategy is more efficient, particularly for temporal and volumetric imaging, than other types of weak annotations, such as per-pixel, bounding boxes, scribbles, ellipses, and points. Furthermore, to exploit our Drag&Drop annotations, we develop a novel weakly supervised learning method based on the watershed algorithm. Experimental results show that our method achieves better detection and localization performance than alternative weak annotations and, more importantly, achieves similar performance to that trained on detailed per-pixel annotations. Interestingly, we find that, with limited resources, allocating weak annotations from a diverse patient population can foster models more robust to unseen images than allocating per-pixel annotations for a small set of images. In summary, this research proposes an efficient annotation strategy for tumor detection and localization that is less accurate than per-pixel annotations but useful for creating large-scale datasets for screening tumors in various medical modalities.

With large language models (LLMs) poised to become embedded in our daily lives, questions are starting to be raised about the dataset(s) they learned from. These questions range from potential bias or misinformation LLMs could retain from their training data to questions of copyright and fair use of human-generated text. However, while these questions emerge, developers of the recent state-of-the-art LLMs become increasingly reluctant to disclose details on their training corpus. We here introduce the task of document-level membership inference for real-world LLMs, i.e. inferring whether the LLM has seen a given document during training or not. First, we propose a procedure for the development and evaluation of document-level membership inference for LLMs by leveraging commonly used data sources for training and the model release date. We then propose a practical, black-box method to predict document-level membership and instantiate it on OpenLLaMA-7B with both books and academic papers. We show our methodology to perform very well, reaching an impressive AUC of 0.856 for books and 0.678 for papers. We then show our approach to outperform the sentence-level membership inference attacks used in the privacy literature for the document-level membership task. We finally evaluate whether smaller models might be less sensitive to document-level inference and show OpenLLaMA-3B to be approximately as sensitive as OpenLLaMA-7B to our approach. Taken together, our results show that accurate document-level membership can be inferred for LLMs, increasing the transparency of technology poised to change our lives.

When deploying machine learning systems to the wild, it is highly desirable for them to effectively leverage prior knowledge to the unfamiliar domain while also firing alarms to anomalous inputs. In order to address these requirements, Universal Domain Adaptation (UniDA) has emerged as a novel research area in computer vision, focusing on achieving both adaptation ability and robustness (i.e., the ability to detect out-of-distribution samples). While UniDA has led significant progress in computer vision, its application on language input still needs to be explored despite its feasibility. In this paper, we propose a comprehensive benchmark for natural language that offers thorough viewpoints of the model's generalizability and robustness. Our benchmark encompasses multiple datasets with varying difficulty levels and characteristics, including temporal shifts and diverse domains. On top of our testbed, we validate existing UniDA methods from computer vision and state-of-the-art domain adaptation techniques from NLP literature, yielding valuable findings: We observe that UniDA methods originally designed for image input can be effectively transferred to the natural language domain while also underscoring the effect of adaptation difficulty in determining the model's performance.

Large language models (LLMs) are transforming the ways the general public accesses and consumes information. Their influence is particularly pronounced in pivotal sectors like healthcare, where lay individuals are increasingly appropriating LLMs as conversational agents for everyday queries. While LLMs demonstrate impressive language understanding and generation proficiencies, concerns regarding their safety remain paramount in these high-stake domains. Moreover, the development of LLMs is disproportionately focused on English. It remains unclear how these LLMs perform in the context of non-English languages, a gap that is critical for ensuring equity in the real-world use of these systems.This paper provides a framework to investigate the effectiveness of LLMs as multi-lingual dialogue systems for healthcare queries. Our empirically-derived framework XlingEval focuses on three fundamental criteria for evaluating LLM responses to naturalistic human-authored health-related questions: correctness, consistency, and verifiability. Through extensive experiments on four major global languages, including English, Spanish, Chinese, and Hindi, spanning three expert-annotated large health Q&A datasets, and through an amalgamation of algorithmic and human-evaluation strategies, we found a pronounced disparity in LLM responses across these languages, indicating a need for enhanced cross-lingual capabilities. We further propose XlingHealth, a cross-lingual benchmark for examining the multilingual capabilities of LLMs in the healthcare context. Our findings underscore the pressing need to bolster the cross-lingual capacities of these models, and to provide an equitable information ecosystem accessible to all.

Prompt Tuning is emerging as a scalable and cost-effective method to fine-tune Pretrained Language Models (PLMs), which are often referred to as Large Language Models (LLMs). This study benchmarks the performance and computational efficiency of Prompt Tuning and baselines for multi-label text classification. This is applied to the challenging task of classifying companies into an investment firm's proprietary industry taxonomy, supporting their thematic investment strategy. Text-to-text classification is frequently reported to outperform task-specific classification heads, but has several limitations when applied to a multi-label classification problem where each label consists of multiple tokens: (a) Generated labels may not match any label in the label taxonomy; (b) The fine-tuning process lacks permutation invariance and is sensitive to the order of the provided labels; (c) The model provides binary decisions rather than appropriate confidence scores. Limitation (a) is addressed by applying constrained decoding using Trie Search, which slightly improves classification performance. All limitations (a), (b), and (c) are addressed by replacing the PLM's language head with a classification head, which is referred to as Prompt Tuned Embedding Classification (PTEC). This improves performance significantly, while also reducing computational costs during inference. In our industrial application, the training data is skewed towards well-known companies. We confirm that the model's performance is consistent across both well-known and less-known companies. Our overall results indicate the continuing need to adapt state-of-the-art methods to domain-specific tasks, even in the era of PLMs with strong generalization abilities. We release our codebase and a benchmarking dataset at https://github.com/EQTPartners/PTEC.

As the complexity of machine learning (ML) models increases and their application in different (and critical) domains grows, there is a strong demand for more interpretable and trustworthy ML. A direct, model-agnostic, way to interpret such models is to train surrogate models, such as rule sets and decision trees, that sufficiently approximate the original ones while being simpler and easier-to-explain. Yet, rule sets can become very lengthy, with many if-else statements, and decision tree depth grows rapidly when accurately emulating complex ML models. In such cases, both approaches can fail to meet their core goal, providing users with model interpretability. To tackle this, we propose DeforestVis, a visual analytics tool that offers user-friendly summarization of the behavior of complex ML models by providing surrogate decision stumps (one-level decision trees) generated with the adaptive boosting (AdaBoost) technique. DeforestVis helps users to explore the complexity vs fidelity trade-off by incrementally generating more stumps, creating attribute-based explanations with weighted stumps to justify decision making, and analyzing the impact of rule overriding on training instance allocation between one or more stumps. An independent test set allows users to monitor the effectiveness of manual rule changes and form hypotheses based on case-by-case analyses. We show the applicability and usefulness of DeforestVis with two use cases and expert interviews with data analysts and model developers.