We introduce an adaptive method with formal quality guarantees for weak supervision in a non-stationary setting. Our goal is to infer the unknown labels of a sequence of data by using weak supervision sources that provide independent noisy signals of the correct classification for each data point. This setting includes crowdsourcing and programmatic weak supervision. We focus on the non-stationary case, where the accuracy of the weak supervision sources can drift over time, e.g., because of changes in the underlying data distribution. Due to the drift, older data could provide misleading information to infer the label of the current data point. Previous work relied on a priori assumptions on the magnitude of the drift to decide how much data to use from the past. In contrast, our algorithm does not require any assumptions on the drift, and it adapts based on the input by dynamically varying its window size. In particular, at each step, our algorithm estimates the current accuracies of the weak supervision sources by identifying a window of past observations that guarantees a near-optimal minimization of the trade-off between the error due to the variance of the estimation and the error due to the drift. Experiments on synthetic and real-world labelers show that our approach adapts to the drift.

This paper presents a comprehensive analysis of the linguistic diversity of LLM safety research, highlighting the English-centric nature of the field.Through a systematic review of nearly 300 publications from 2020-2024 across major NLP conferences and workshops at * ACL, we identify a significant and growing language gap in LLM safety research, with even high-resource non-English languages receiving minimal attention.We further observe that non-English languages are rarely studied as a standalone language and that English safety research exhibits poor language documentation practice.To motivate future research into multilingual safety, we make several recommendations based on our survey, and we then pose three concrete future directions on safety evaluation, training data generation, and crosslingual safety generalization.Based on our survey and proposed directions, the field can develop more robust, inclusive AI safety practices for diverse global populations.

Biomedical knowledge graphs (KGs) encode rich, structured information critical for drug discovery tasks, but extracting meaningful insights from large-scale KGs remains challenging due to their complex structure. Existing biomedical subgraph retrieval methods are tailored for graph neural networks (GNNs), limiting compatibility with other paradigms, including large language models (LLMs). We introduce K-Paths, a model-agnostic retrieval framework that extracts structured, diverse, and biologically meaningful multi-hop paths from dense biomedical KGs. These paths enable prediction of unobserved drug-drug and drug-disease interactions, including those involving entities not seen during training, thus supporting inductive reasoning. K-Paths is training-free and employs a diversity-aware adaptation of Yen's algorithm to extract the K shortest loopless paths between entities in a query, prioritizing biologically relevant and relationally diverse connections. These paths serve as concise, interpretable reasoning chains that can be directly integrated with LLMs or GNNs to improve generalization, accuracy, and enable explainable inference. Experiments on benchmark datasets show that K-Paths improves zero-shot reasoning across state-of-the-art LLMs. For instance, Tx-Gemma 27B improves by 19.8 and 4.0 F1 points on interaction severity prediction and drug repurposing tasks, respectively. Llama 70B achieves gains of 8.5 and 6.2 points on the same tasks. K-Paths also boosts the training efficiency of EmerGNN, a state-of-the-art GNN, by reducing the KG size by 90% while maintaining predictive performance. Beyond efficiency, K-Paths bridges the gap between KGs and LLMs, enabling scalable and explainable LLM-augmented scientific discovery. We release our code and the retrieved paths as a benchmark for inductive reasoning.

Max Zuo, Francisco Piedrahita Velez, Xiaochen Li, Michael Littman, Stephen Bach. Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers). 2025.

Recent works often assume that Vision-Language Model (VLM) representations are based on visual attributes like shape.However, it is unclear to what extent VLMs prioritize this information to represent concepts.We propose Extract and Explore (EX2), a novel approach to characterize textual features that are important for VLMs.EX2 uses reinforcement learning to align a large language model with VLM preferences and generates descriptions that incorporate features that are important for the VLM.Then, we inspect the descriptions to identify features that contribute to VLM representations.Using EX2, we find that spurious descriptions have a major role in VLM representations despite providing no helpful information, e.g., Click to enlarge photo of CONCEPT.More importantly, among informative descriptions, VLMs rely significantly on non-visual attributes like habitat (e.g., North America) to represent visual concepts.Also, our analysis reveals that different VLMs prioritize different attributes in their representations.Overall, we show that VLMs do not simply match images to scene descriptions and that non-visual or even spurious descriptions significantly influence their representations.

A promising approach for improving the performance of vision-language models like CLIP for image classification is to extend the class descriptions (i.e., prompts) with related attributes, e.g., using brown sparrow instead of sparrow. However, current zero-shot methods select a subset of attributes regardless of commonalities between the target classes, potentially providing no useful information that would have helped to distinguish between them. For instance, they may use color instead of bill shape to distinguish between sparrows and wrens, which are both brown. We propose Follow-up Differential Descriptions (FuDD), a zero-shot approach that tailors the class descriptions to each dataset and leads to additional attributes that better differentiate the target classes. FuDD first identifies the ambiguous classes for each image, and then uses a Large Language Model (LLM) to generate new class descriptions that differentiate between them. The new class descriptions resolve the initial ambiguity and help predict the correct label. In our experiments, FuDD consistently outperforms generic description ensembles and naive LLM-generated descriptions on 12 datasets. We show that differential descriptions are an effective tool to resolve class ambiguities, which otherwise significantly degrade the performance. We also show that high quality natural language class descriptions produced by FuDD result in comparable performance to few-shot adaptation methods.

Prompted weak supervision (PromptedWS) applies pre-trained large language models (LLMs) as the basis for labeling functions (LFs) in a weak supervision framework to obtain large labeled datasets. We further extend the use of LLMs in the loop to address one of the key challenges in weak supervision: learning the statistical dependency structure among supervision sources. In this work, we ask the LLM how similar are these prompted LFs. We propose a Structure Refining Module, a simple yet effective first approach based on the similarities of the prompts by taking advantage of the intrinsic structure in the embedding space. At the core of Structure Refining Module are Labeling Function Removal (LaRe) and Correlation Structure Generation (CosGen). Compared to previous methods that learn the dependencies from weak labels, our method finds the dependencies which are intrinsic to the LFs and less dependent on the data. We show that our Structure Refining Module improves the PromptedWS pipeline by up to 12.7 points on the benchmark tasks. We also explore the trade-offs between efficiency and performance with comprehensive ablation experiments and analysis. Code for this project can be found in https://github.com/BatsResearch/su-bigdata23-code.

Fine-tuning vision-language models (VLMs) like CLIP to downstream tasks is often necessary to optimize their performance. However, a major obstacle is the limited availability of labeled data. We study the use of pseudolabels, i.e., heuristic labels for unlabeled data, to enhance CLIP via prompt tuning. Conventional pseudolabeling trains a model on labeled data and then generates labels for unlabeled data. VLMs' zero-shot capabilities enable a ``second generation'' of pseudolabeling approaches that do not require task-specific training on labeled data. By using zero-shot pseudolabels as a source of supervision, we observe that learning paradigms such as semi-supervised, transductive zero-shot, and unsupervised learning can all be seen as optimizing the same loss function. This unified view enables the development of versatile training strategies that are applicable across learning paradigms. We investigate them on image classification tasks where CLIP exhibits limitations, by varying prompt modalities, e.g., textual or visual prompts, and learning paradigms. We find that (1) unexplored prompt tuning strategies that iteratively refine pseudolabels consistently improve CLIP accuracy, by 19.5 points in semi-supervised learning, by 28.4 points in transductive zero-shot learning, and by 15.2 points in unsupervised learning, and (2) unlike conventional semi-supervised pseudolabeling, which exacerbates model biases toward classes with higher-quality pseudolabels, prompt tuning leads to a more equitable distribution of per-class accuracy. The code to reproduce the experiments is at https://github.com/BatsResearch/menghini-neurips23-code.

As post hoc explanation methods are increasingly being leveraged to explain complex models in high-stakes settings, it becomes critical to ensure that the quality of the resulting explanations is consistently high across all subgroups of a population. For instance, it should not be the case that explanations associated with instances belonging to, e.g., women, are less accurate than those associated with other genders. In this work, we initiate the study of identifying group-based disparities in explanation quality. To this end, we first outline several key properties that contribute to explanation quality-namely, fidelity (accuracy), stability, consistency, and sparsity-and discuss why and how disparities in these properties can be particularly problematic. We then propose an evaluation framework which can quantitatively measure disparities in the quality of explanations. Using this framework, we carry out an empirical analysis with three datasets, six post hoc explanation methods, and different model classes to understand if and when group-based disparities in explanation quality arise. Our results indicate that such disparities are more likely to occur when the models being explained are complex and non-linear. We also observe that certain post hoc explanation methods (e.g., Integrated Gradients, SHAP) are more likely to exhibit disparities. Our work sheds light on previously unexplored ways in which explanation methods may introduce unfairness in real world decision making.

Programmatic weak supervision creates models without hand-labeled training data by combining the outputs of heuristic labelers. Existing frameworks make the restrictive assumption that labelers output a single class label. Enabling users to create partial labelers that output subsets of possible class labels would greatly expand the expressivity of programmatic weak supervision. We introduce this capability by defining a probabilistic generative model that can estimate the underlying accuracies of multiple noisy partial labelers without ground truth labels. We show how to scale up learning, for example learning on 100k examples in one minute, a 300$\times$ speed up compared to a naive implementation. We also prove that this class of models is generically identifiable up to label swapping under mild conditions. We evaluate our framework on three text classification and six object classification tasks. On text tasks, adding partial labels increases average accuracy by 8.6 percentage points. On image tasks, we show that partial labels allow us to approach some zero-shot object classification problems with programmatic weak supervision by using class attributes as partial labelers. On these tasks, our framework has accuracy comparable to recent embedding-based zero-shot learning methods, while using only pre-trained attribute detectors.

Training and evaluating language models increasingly requires the construction of meta-datasets -- diverse collections of curated data with clear provenance. Natural language prompting has recently lead to improved zero-shot generalization by transforming existing, supervised datasets into a variety of novel instruction tuning tasks, highlighting the benefits of meta-dataset curation. While successful in general-domain text, translating these data-centric approaches to biomedical language modeling remains challenging, as labeled biomedical datasets are significantly underrepresented in popular data hubs. To address this challenge, we introduce BigBio a community library of 126+ biomedical NLP datasets, currently covering 13 task categories and 10+ languages. BigBio facilitates reproducible meta-dataset curation via programmatic access to datasets and their metadata, and is compatible with current platforms for prompt engineering and end-to-end few/zero shot language model evaluation. We discuss our process for task schema harmonization, data auditing, contribution guidelines, and outline two illustrative use cases: zero-shot evaluation of biomedical prompts and large-scale, multi-task learning. BigBio is an ongoing community effort and is available at https://github.com/bigscience-workshop/biomedical

We develop a rigorous mathematical analysis of zero-shot learning with attributes. In this setting, the goal is to label novel classes with no training data, only detectors for attributes and a description of how those attributes are correlated with the target classes, called the class-attribute matrix. We develop the first non-trivial lower bound on the worst-case error of the best map from attributes to classes for this setting, even with perfect attribute detectors. The lower bound characterizes the theoretical intrinsic difficulty of the zero-shot problem based on the available information---the class-attribute matrix---and the bound is practically computable from it. Our lower bound is tight, as we show that we can always find a randomized map from attributes to classes whose expected error is upper bounded by the value of the lower bound. We show that our analysis can be predictive of how standard zero-shot methods behave in practice, including which classes will likely be confused with others.

We develop a novel method that provides theoretical guarantees for learning from weak labelers without the (mostly unrealistic) assumption that the errors of the weak labelers are independent or come from a particular family of distributions. We show a rigorous technique for efficiently selecting small subsets of the labelers so that a majority vote from such subsets has a provably low error rate. We explore several extensions of this method and provide experimental results over a range of labeled data set sizes on 45 image classification tasks. Our performance-guaranteed methods consistently match the best performing alternative, which varies based on problem difficulty. On tasks with accurate weak labelers, our methods are on average 3 percentage points more accurate than the state-of-the-art adversarial method. On tasks with inaccurate weak labelers, our methods are on average 15 percentage points more accurate than the semi-supervised Dawid-Skene model (which assumes independence).

We develop a rigorous approach for using a set of arbitrarily correlated weak supervision sources in order to solve a multiclass classification task when only a very small set of labeled data is available. Our learning algorithm provably converges to a model that has minimum empirical risk with respect to an adversarial choice over feasible labelings for a set of unlabeled data, where the feasibility of a labeling is computed through constraints defined by rigorously estimated statistics of the weak supervision sources. We show theoretical guarantees for this approach that depend on the information provided by the weak supervision sources. Notably, this method does not require the weak supervision sources to have the same labeling space as the multiclass classification task. We demonstrate the effectiveness of our approach with experiments on various image classification tasks.

We propose a framework for training sequence tagging models with weak supervision consisting of multiple heuristic rules of unknown accuracy. In addition to supporting rules that vote on tags in the output sequence, we introduce a new type of weak supervision, called linking rules, that vote on how sequence elements should be grouped into spans with the same tag. These rules are an alternative to candidate span generators that require significantly more human effort. To estimate the accuracies of the rules and combine their conflicting outputs into training data, we introduce a new type of generative model, linked hidden Markov models (linked HMMs), and prove they are generically identifiable (up to a tag permutation) without any observed training labels. We find that linked HMMs provide an average 7 F1 point boost on benchmark named entity recognition tasks versus generative models that assume the tags are i.i.d. Further, neural sequence taggers trained with these structure-aware generative models outperform comparable state-of-the-art approaches to weak supervision by an average of 2.6 F1 points.

Labeling training data is increasingly the largest bottleneck in deploying machine learning systems. We present Snorkel, a first-of-its-kind system that enables users to train state-of-the-art models without hand labeling any training data. Instead, users write labeling functions that express arbitrary heuristics, which can have unknown accuracies and correlations. Snorkel denoises their outputs without access to ground truth by incorporating the first end-to-end implementation of our recently proposed machine learning paradigm, data programming. We present a flexible interface layer for writing labeling functions based on our experience over the past year collaborating with companies, agencies, and research laboratories. In a user study, subject matter experts build models $$2.8\times $$ faster and increase predictive performance an average $$45.5\%$$ versus seven hours of hand labeling. We study the modeling trade-offs in this new setting and propose an optimizer for automating trade-off decisions that gives up to $$1.8\times $$ speedup per pipeline execution. In two collaborations, with the US Department of Veterans Affairs and the US Food and Drug Administration, and on four open-source text and image data sets representative of other deployments, Snorkel provides $$132\%$$ average improvements to predictive performance over prior heuristic approaches and comes within an average $$3.60\%$$ of the predictive performance of large hand-curated training sets.

Labeling training data is one of the most costly bottlenecks in developing machine learning-based applications. We present a first-of-its-kind study showing how existing knowledge resources from across an organization can be used as weak supervision in order to bring development time and cost down by an order of magnitude, and introduce Snorkel DryBell, a new weak supervision management system for this setting. Snorkel DryBell builds on the Snorkel framework, extending it in three critical aspects: flexible, template-based ingestion of diverse organizational knowledge, cross-feature production serving, and scalable, sampling-free execution. On three classification tasks at Google, we find that Snorkel DryBell creates classifiers of comparable quality to ones trained with tens of thousands of hand-labeled examples, converts non-servable organizational resources to servable models for an average 52% performance improvement, and executes over millions of data points in tens of minutes.

Curating labeled training data has become the primary bottleneck in machine learning. Recent frameworks address this bottleneck with generative models to synthesize labels at scale from weak supervision sources. The generative model’s dependency structure directly affects the quality of the estimated labels, but selecting a structure automatically without any labeled data is a distinct challenge. We propose a structure estimation method that maximizes the l1-regularized marginal pseudolikelihood of the observed data. Our analysis shows that the amount of unlabeled data required to identify the true structure scales sublinearly in the number of possible dependencies for a broad class of models. Simulations show that our method is 100x faster than a maximum likelihood approach and selects 1/4 as many extraneous dependencies. We also show that our method provides an average of 1.5 F1 points of improvement over existing, user-developed information extraction applications on real-world data such as PubMed journal abstracts.

State-of-the art machine learning methods such as deep learning rely on large sets of hand-labeled training data. Collecting training data is prohibitively slow and expensive, especially when technical domain expertise is required; even the largest technology companies struggle with this challenge. We address this critical bottleneck with Snorkel, a new system for quickly creating, managing, and modeling training sets. Snorkel enables users to generate large volumes of training data by writing labeling functions, which are simple functions that express heuristics and other weak supervision strategies. These user-authored labeling functions may have low accuracies and may overlap and conflict, but Snorkel automatically learns their accuracies and synthesizes their output labels. Experiments and theory show that surprisingly, by modeling the labeling process in this way, we can train high-accuracy machine learning models even using potentially lower-accuracy inputs. Snorkel is currently used in production at top technology and consulting companies, and used by researchers to extract information from electronic health records, after-action combat reports, and the scientific literature. In this demonstration, we focus on the challenging task of information extraction, a common application of Snorkel in practice. Using the task of extracting corporate employment relationships from news articles, we will demonstrate and build intuition for a radically different way of developing machine learning systems which allows us to effectively bypass the bottleneck of hand-labeling training data.

A fundamental challenge in developing high-impact machine learning technologies is balancing the need to model rich, structured domains with the ability to scale to big data. Many important problem areas are both richly structured and large scale, from social and biological networks, to knowledge graphs and the Web, to images, video, and natural language. In this paper, we introduce two new formalisms for modeling structured data, and show that they can both capture rich structure and scale to big data. The first, hinge-loss Markov random fields (HL-MRFs), is a new kind of probabilistic graphical model that generalizes different approaches to convex inference. We unite three approaches from the randomized algorithms, probabilistic graphical models, and fuzzy logic communities, showing that all three lead to the same inference objective. We then define HL- MRFs by generalizing this unified objective. The second new formalism, probabilistic soft logic (PSL), is a probabilistic programming language that makes HL-MRFs easy to define using a syntax based on first-order logic. We introduce an algorithm for inferring most-probable variable assignments (MAP inference) that is much more scalable than general-purpose convex optimization methods, because it uses message passing to take advantage of sparse dependency structures. We then show how to learn the parameters of HL-MRFs. The learned HL-MRFs are as accurate as analogous discrete models, but much more scalable. Together, these algorithms enable HL-MRFs and PSL to model rich, structured data at scales not previously possible.

Labeling training data is increasingly the largest bottleneck in deploying machine learning systems. We present Snorkel, a first-of-its-kind system that enables users to train state-of- the-art models without hand labeling any training data. Instead, users write labeling functions that express arbitrary heuristics, which can have unknown accuracies and correlations. Snorkel denoises their outputs without access to ground truth by incorporating the first end-to-end implementation of our recently proposed machine learning paradigm, data programming. We present a flexible interface layer for writing labeling functions based on our experience over the past year collaborating with companies, agencies, and research labs. In a user study, subject matter experts build models 2.8× faster and increase predictive performance an average 45.5% versus seven hours of hand labeling. We study the modeling tradeoffs in this new setting and propose an optimizer for automating tradeoff decisions that gives up to 1.8× speedup per pipeline execution. In two collaborations, with the U.S. Department of Veterans Affairs and the U.S. Food and Drug Administration, and on four open-source text and image data sets representative of other deployments, Snorkel provides 132% average improvements to predictive performance over prior heuristic approaches and comes within an average 3.60% of the predictive performance of large hand-curated training sets.

One of the most important obstacles to deploying predictive models is the fact that humans do not understand and trust them. Knowing which variables are important in a model's prediction and how they are combined can be very powerful in helping people understand and trust automatic decision making systems. Here we propose interpretable decision sets, a framework for building predictive models that are highly accurate, yet also highly interpretable. Decision sets are sets of independent if-then rules. Because each rule can be applied independently, decision sets are simple, concise, and easily interpretable. We formalize decision set learning through an objective function that simultaneously optimizes accuracy and interpretability of the rules. In particular, our approach learns short, accurate, and non-overlapping rules that cover the whole feature space and pay attention to small but important classes. Moreover, we prove that our objective is a non-monotone submodular function, which we efficiently optimize to find a near-optimal set of rules. Experiments show that interpretable decision sets are as accurate at classification as state-of-the-art machine learning techniques. They are also three times smaller on average than rule-based models learned by other methods. Finally, results of a user study show that people are able to answer multiple-choice questions about the decision boundaries of interpretable decision sets and write descriptions of classes based on them faster and more accurately than with other rule-based models that were designed for interpretability. Overall, our framework provides a new approach to interpretable machine learning that balances accuracy, interpretability, and computational efficiency.

We prove the equivalence of first-order local consistency relaxations and the MAX SAT relaxation of Goemans and Williamson (1994) for a class of MRFs we refer to as logical MRFs. This allows us to combine the advantages of each into a single MAP inference technique: solving the local consistency relaxation with any of a number of highly scalable message-passing algorithms, and then obtaining a high-quality discrete solution via a guaranteed rounding procedure when the relaxation is not tight. Logical MRFs are a general class of models that can incorporate many common dependencies, such as logical implications and mixtures of supermodular and submodular potentials. They can be used for many structured prediction tasks, including natural language processing, computer vision, and computational social science. We show that our new inference technique can improve solution quality by as much as 20% without sacrificing speed on problems with over one million dependencies.

Latent variables allow probabilistic graphical models to capture nuance and structure in important domains such as network science, natural language processing, and computer vision. Naive approaches to learning such complex models can be prohibitively expensive—because they require repeated inferences to update beliefs about latent variables—so lifting this restriction for useful classes of models is an important problem. Hinge-loss Markov random fields (HL-MRFs) are graphical models that allow highly scalable inference and learning in structured domains, in part by representing structured problems with continuous variables. However, this representation leads to challenges when learning with latent variables. We introduce paired-dual learning, a framework that greatly speeds up training by using tractable entropy surrogates and avoiding repeated inferences. Paired-dual learning optimizes an objective with a pair of dual inference problems. This allows fast, joint optimization of parameters and dual variables. We evaluate on social-group detection, trust prediction in social networks, and image reconstruction, finding that paired-dual learning trains models as accurate as those trained by traditional methods in much less time, often before traditional methods make even a single parameter update.

Probabilistic graphical models are powerful tools for analyzing constrained, continuous domains. However, finding most-probable explanations (MPEs) in these models can be computationally expensive. In this paper, we improve the scalability of MPE inference in a class of graphical models with piecewise-linear and piecewise-quadratic dependencies and linear constraints over continuous domains. We derive algorithms based on a consensus-optimization framework and demonstrate their superior performance over state of the art. We show empirically that in a large-scale voter-preference modeling problem our algorithms scale linearly in the number of dependencies and constraints.

To cope with concept drift, we placed a probability distribution over the location of the most-recent drift point. We used Bayesian model comparison to update this distribution from the predictions of models trained on blocks of consecutive observations and pruned potential drift points with low probability. We compare our approach to a non-probabilistic method for drift and a probabilistic method for change-point detection. In our experiments, our approach generally yielded improved accuracy and/or speed over these other methods.

To cope with concept drift, we paired a stable online learner with a reactive one. A stable learner predicts based on all of its experience, whereas are active learner predicts based on its experience over a short, recent window of time. The method of paired learning uses differences in accuracy between the two learners over this window to determine when to replace the current stable learner, since the stable learner performs worse than does there active learner when the target concept changes. While the method uses the reactive learner as an indicator of drift, it uses the stable learner to predict, since the stable learner performs better than does the reactive learner when acquiring target concept. Experimental results support these assertions. We evaluated the method by making direct comparisons to dynamic weighted majority, accuracy weighted ensemble, and streaming ensemble algorithm (SEA) using two synthetic problems, the Stagger concepts and the SEA concepts, and three real-world data sets: meeting scheduling, electricity prediction, and malware detection. Results suggest that, on these problems, paired learners outperformed or performed comparably to methods more costly in time and space.