论文检索
输入标题、作者或关键词，从 2,854 篇学术成果中精准定位
The increasing use of neural networks (NNs) in high-stakes decision-making requires rigorous analysis to ensure safety, fairness, and explainability. Formal verification tools for neural networks typically focus on determining the satisfiability of properties such as safety or fairness. However, many fairness and explainability tasks go beyond satisfiability and instead rely on arbitrary optimization objectives. To address such problems, neural networks are often encoded as mixed-integer linear optimization (MILO) problems with linear objectives. In practice, these encodings are usually implemented in an ad-hoc manner, limiting comparability across works, reducing transparency, and increasing implementation effort. We address this gap by introducing eNNcode, a user-friendly PyPI library that converts any piecewise-linear neural network in Open Neural Network Exchange (ONNX) format into a MILO instance. The library supports arbitrary constraints on input and output nodes, as well as user-defined optimization objectives. Our experiments show that eNNcode achieves performance comparable to existing libraries, despite its simplicity and ease of use. Overall, eNNcode facilitates reproducible and standardized optimization-based analysis of neural networks.
The literature on AI fairness has grown rapidly, proposing a large number of bias mitigation techniques that are commonly organized into pre-, in-, and post-processing methods. This pipeline-centric view offers an operational, lifecycle-based perspective on where mitigation can be applied. In deployment settings, however, practitioners also face an additional question: whether a mitigation family is applicable given the resources and access rights available in a concrete system. In this survey, we use resources broadly to denote data access/control, training capability, and deployment-time interface/decision control. Accordingly, we introduce a resource-aware taxonomy that complements existing taxonomies by classifying AI bias mitigation methods according to the conditions that make them practically implementable. We use this taxonomy to structure and reinterpret existing literature on the topic, highlighting which mitigation families remain feasible under resource constraints.
Large Language Models (LLMs) have become central to modern AI, with interpretability playing an increasingly important role in investigating the opaque and complex mechanisms encoded within billions of parameters and ensuring trustworthy deployment. However, descriptive interpretability approaches for LLMs remain largely post-hoc, explaining model behavior without actionable guidance for model improvement, thereby limiting their practical utility. Recent work has therefore shifted toward actionable LLM interpretability, emphasizing methods that connect internal mechanisms to interventions and model improvement rather than explanation alone. This survey reviews LLM interpretability through the lens of actionability, presenting a taxonomy of attributional and mechanistic approaches, along with emerging methods tailored to vision–language models (VLMs). We further examine how actionable interpretability supports downstream objectives such as hallucination mitigation, knowledge editing, fairness, safety, capability and efficiency. By positioning actionable interpretability as a pathway for better-guided LLM design and practice, this survey outlines key challenges and future directions toward trustworthy and controllable foundation models.
We present an industrial case on workforce allocation and scheduling in the aircraft manufacturing industry, where available teams need to be assigned to logistical operations. This application presents several challenges, such as the scale of the problem, the need for fair workload distribution, and the need for methods to mitigate unforeseen disruptions due to technical malfunctions or incompatible weather conditions. We compare different Constraint Programming (CP) models for the allocation and scheduling problems, with extra focus on modeling the workload balancing component. Additionally, we investigate automatic rescheduling methods for restoring feasibility after a disruption invalidates the precomputed schedule. Our results show that by using appropriate modeling techniques, the problem can be solved in reasonable time, thereby producing fair schedules. Additionally, we show how invalidated schedules can be restored efficiently to help human operators in resolving disruptions to the schedule.
Fairness in machine learning (ML) has garnered significant attention. However, current research has mainly concentrated on the distributive fairness of ML models, with limited focus on another dimension of fairness, i.e., procedural fairness. In this paper, we first define the procedural fairness of ML models by drawing from the established understanding of procedural fairness in philosophy and psychology fields, and then give formal definitions of individual and group procedural fairness. Based on the proposed definition, we further propose a novel metric to evaluate the group procedural fairness of ML models, called GPFFAE, which utilizes a widely used explainable artificial intelligence technique, namely feature attribution explanation (FAE), to capture the decision process of ML models. We validate the effectiveness of GPFFAE on a synthetic dataset and eight real-world datasets. Our experimental studies have revealed the relationship between procedural and distributive fairness of ML models. After validating the proposed metric for assessing the procedural fairness of ML models, we then propose a method for identifying the features that lead to the procedural unfairness of the model and propose two methods to improve procedural fairness based on the identified unfair features. Our experimental results demonstrate that we can accurately identify the features that lead to procedural unfairness in the ML model, and both of our proposed methods can significantly improve procedural fairness while also improving distributive fairness, with a slight sacrifice on the model performance
We study the fair and truthful allocation of m divisible public items among n agents, each with distinct preferences for the items. To aggregate agents' preferences fairly, we focus on finding a core solution. For divisible items, a core solution always exists and can be calculated by maximizing the Nash welfare objective. However, such a solution is easily manipulated; agents might have incentives to misreport their preferences. To mitigate this, the current state-of-the-art finds an approximate core solution with high probability while ensuring approximate truthfulness. However, this approach has two main limitations. First, due to several approximations, the approximation error in the core could grow with n, resulting in a non-asymptotic core solution. This limitation is particularly significant as public-good allocation mechanisms are frequently applied in scenarios involving a large number of agents, such as the allocation of public tax funds for municipal projects. Second, implementing the current approach for practical applications proves to be a highly nontrivial task. To address these limitations, we introduce PPGA, a (differentially) Private Public-Good Allocation algorithm, and show that it attains asymptotic truthfulness and finds an asymptotic core solution with high probability. Additionally, to demonstrate the practical applicability of our algorithm, we implement PPGA and empirically study its properties using municipal participatory budgeting data.
Many high-stakes social applications of AI, such as public health surveillance and policy planning, operate at the community- rather than individual-level. However, most AI fairness research operates at the individual- or data-level (i.e. document or image) and rely on metrics defined over discrete demographic categories rather than population-level demographic proportions. In this work, we first introduce the Bilateral Concentration Index (BCI) to quantify nonmonotonic error disparities missed by the category-based metrics used at individual or data-levels. Then we conduct a large-scale audit of sociodemographic error disparities in both lexical- and transformer-based models of county-level health outcomes, over a dataset cover billions of community-mapped messages. While all tasks had significant disparity, the size varied widely depending on the outcome and model, from BCI of 2.1% for predicting life satisfaction to 17.0% for predicting fair or poor health. We further evaluate four approaches for incorporating sociodemographic information, as potential bias mitigation strategies, finding that while demographic inclusion consistently improved predictive accuracy, it frequently amplified error disparities. The largest disparities were associated with education and income (BCI = 2.7–16.4%), often reducing accuracy for low-income, and in some cases, high-income communities. These findings highlight a critical accuracy–fairness trade-off in community-level models for public health tasks, demonstrating how seemingly beneficial modeling choices can lead to increased disparities which could disadvantage communities if model predictions are used to make policy decisions in geographic areas where this public health data is not available.
Recent research shows that conversational AI can shift voter preferences, with effects persisting for weeks. Yet frontier models exhibit a documented "persuasion-reliability tradeoff", producing hallucinated or systematically distorted election information. Despite these risks, election officials lack standardized tools to systematically evaluate AI systems before deployment. We propose CivicAudit-Bench, a stakeholder-guided auditing framework to stress-test large language models for civic hallucinations, false confidence, jurisdiction-dependent failure, and asymmetric refusals/accuracy. This framework introduces a modular, counterfactual, and severity-aware auditing methodology that integrates roll-call–based alignment modeling, entity-swap probing, and jurisdiction-conditional correctness criteria. Informed by engagement with the U.S. Election Assistance Commission, the toolkit consists of three modules: (1) PoliBias-US, a multi-indicator alignment screen combining Congressional roll-call ideology scaling with party-cue counterfactual sensitivity, persona robustness, and narrative-framing alignment; (2) HalluBias-Election, an evidence-linked benchmark that measures hallucinations, severity-weighted critical errors, and asymmetries via Entity-Swap Counterfactual Probing and a jurisdiction-safe completion criterion; and (3) Disclosure-Test, pre-registered experiments assessing whether transparency and calibrated-uncertainty disclosures reduce overreliance and attenuate persuasion without blocking legitimate civic information. CivicAudit-Bench outputs versioned audit scorecards and a coordinated white-hat disclosure workflow, advancing UN SDG~16 by strengthening democratic information integrity.
Motivated by the problem of assigning mediators to cases in the Kenyan judicial system, we study an online resource allocation problem where incoming tasks (cases) must be immediately assigned to available, capacity-constrained resources (mediators). The resources differ in their quality, which may need to be learned. In addition, resources can only be assigned to a subset of tasks that overlaps to varying degrees with the subset of tasks other resources can be assigned to. The objective is to maximize task completion while satisfying soft capacity constraints across all the resources. The scale of the real-world problem poses substantial challenges, since there are over 2000 mediators, and a multitude of combinations of geographic locations (87) and case types (12) that each mediator is qualified to work on. Together, these features—unknown quality of new resources (newly onboarded mediators), soft capacity constraints (due to the mandate to assign cases without delay), and high-dimensional state space—make existing scheduling and resource allocation algorithms either inapplicable or inefficient. We formalize the problem in a tractable manner, using a quadratic program formulation for assignment and a multi-agent bandit style framework for learning. We demonstrate the key properties and advantages of our new algorithm, SMaRT (Selecting Mediators that are Right for the Task), compared with baselines on some stylized instances of the mediator allocation problem. We then turn to considering its application to real-world data on cases and mediators from the Kenyan Judiciary. SMaRT outperforms baselines and allows for controlling the tradeoff between the strictness of the capacity constraints and overall case resolution rates, both in situations where mediator quality is known beforehand and when the problem is bandit-like in that learning is part of the problem definition. On the strength of these results, we plan to conduct a randomized controlled trial in which we deploy SMaRT in the Judiciary's mediation management system.
Multimodal meme understanding is increasingly used to analyze socially sensitive content, yet existing models often exhibit biased behavior when interpreting economic dependence and social roles under ambiguity. Many memes express economic relationships through sparse text or symbolic visual cues, providing insufficient evidence for gendered attribution. In such underspecified settings, models tend to rely on pretraining correlations, leading to hallucinated and stereotypical economic role assignments. In this work, we study gendered economic dependence in image-text memes through the lens of contextual sufficiency and identify epistemic overcommitment—inferring roles without adequate evidence—as a primary source of bias. We propose CGER-Net, a context-grounded multimodal framework that estimates whether the input provides sufficient evidence for gendered economic reasoning and applies evidence-gated inference to enable confident attribution when cues are explicit while favoring principled abstention otherwise. We evaluate CGER-Net on EconMeme-GE, a curated dataset of image-text memes annotated as Men, Women, Neutral, or Ambiguous. Across strong contemporary multimodal baselines, CGER-Net reduces Gender Overcommitment Rate by up to 44% on ambiguous instances while maintaining comparable accuracy on unambiguous cases. Human evaluation further shows that 79% of generated rationales are judged as epistemically aligned with the available evidence. These results highlight the importance of modeling when not to infer for reliable and responsible multimodal analysis.
Climate policy analysis requires models that capture multi-gas climate effects, but such models are too slow to embed in reinforcement learning loops at scale. In collaboration with a pan-European public-sector environmental agency, we develop a multi-agent reinforcement learning (MARL) framework that integrates a higher-fidelity climate surrogate as the environment transition, enabling regional agents to learn policies under multi-gas dynamics. We train a recurrent surrogate on 20,000 multi-gas emission pathways to emulate CICERO-SCM. The surrogate achieves near-simulator accuracy (global-mean temperature RMSE ≈ 4 × 10⁻⁴ K) with ∼1000× faster one-step inference and yields >100× end-to-end MARL training speed-up. We show policy agreement with the simulator in tractable settings and propose a replay- and rank-consistency test (Kendall’s τ) for assessing policy fidelity when simulator-in-the-loop training is infeasible. This enables large-scale multi-agent policy experiments while retaining high-fidelity multi-gas climate response.
Large language models (LLMs) are increasingly used to answer questions about political information, including in election-adjacent information settings where factual errors and ideological distortions are high-stakes. We present a reproducible measurement framework that treats hallucinations, unsupported statements in document-grounded QA, as diagnostic signals of ideological drift. Using 21,727 expert-labeled U.S. political news articles from QBias spanning left, center, and right sources, we (i) generate an article-specific question, (ii) elicit document-grounded answers from three open-weight LLMs and one proprietary model, (iii) detect sentence-level hallucinations via reference-based comparison, (iv) classify the ideological valence of hallucinated sentences with a fine-tuned stance classifier, and (v) probe output logits to relate token-level uncertainty to hallucination and drift. Hallucination rates vary substantially across models and concentrate in contentious topics, while source-ideology differences in hallucination frequency are modest. In contrast, hallucination content exhibits robust leftward drift: a majority of hallucinated sentences are classified as left-leaning, including among hallucinations generated from right-leaning sources. Logit-level analysis shows hallucinations arise in high-entropy generation contexts, and in some models uncertainty also predicts leftward drift, consistent with an "uncertainty → guessing" mechanism. We discuss implications for auditing AI-mediated political information and for designing safeguards in election-relevant deployments.
Every day, many people die under violent circumstances, whether from crimes, war, migration, or climate disasters. Medico-legal and law enforcement institutions document many portraits of the deceased for evidence, but cannot immediately carry out identification on them. While traditional image editing tools can process these photos for public release, the workflow is lengthy and produces suboptimal results. In this work, we leverage advances in image generation models, which can now produce photorealistic human portraits, to introduce FlowID, an identity-preserving facial reconstruction method. Our approach combines single-image fine-tuning, which adapts the generative model to out-of-distribution injured faces, with attention-based masking that localizes edits to damaged regions while preserving identity-critical features. Together, these components enable the removal of artifacts from violent death while retaining sufficient identity information to support identification. To evaluate our method, we introduce InjuredFaces, a novel benchmark for identity-preserving facial reconstruction under severe facial damage. Beyond serving as an evaluation tool for this work, InjuredFaces provides a standardized resource for the community to study and compare methods addressing facial reconstruction in extreme conditions. Experimental results show that FlowID outperforms state-of-the-art open-source methods while maintaining low memory requirements, making it suitable for local deployment without compromising data privacy.
The proliferation of sophisticated telephone scams poses a significant societal and economic threat, impacting diverse linguistic contexts in a country like India. Furthermore, the lack of large-scale, publicly available datasets remains a critical barrier impacting research on robust, real-time countermeasures. In view of this, the proposed work introduces ICFD-31k, the first Indian Conversational Fraud Dataset, representing a new benchmark containing over 31,000 realistic conversational transcripts. ICFD-31k comprises systematically generated content, covering 10 distinct fraud umbrellas spanning from financial impersonation to job scams. ICFD-31k transcripts feature rich annotations comprising a final verdict, chunk-level streaming labels, and detailed ``slow-thinking'' rationales. In addition, the human-in-the-loop evaluation validates the ICFD-31k's quality, achieving a Cohen's Kappa of 0.534 that confirms annotation reliability. Furthermore, the proposed work introduces two fine-tuned models based on RoBERTa: M1 for non-streaming data and M2 for streaming data. The comprehensive experiments with strong baselines (M1, M2) further demonstrate the ICFD-31k's utility. The code and reproducibility materials are available at https://github.com/SPELLAILab/ICFD-31k.
Brazil is home to about 180 Indigenous languages, which span a wide range of sociolinguistic circumstances, from critically endangered to relatively stable. Across this continuum, Indigenous communities often lack pedagogical resources and are underserved by existing language-learning technologies, which are typically designed for high-resource languages and assume solid connectivity and large datasets. This research project proposes AI-assisted tools and workflows for the collection and annotation of textual and speech data that substantially reduce the time and cost required to produce engaging language-learning game apps. Our goal is to implement a language-learning game app that Indigenous students can use to practice their reading, writing and speaking skills at home. We propose novel speech processing models for low-resource Indigenous languages and offlline support in low-connectivity environments. Our project adopts a co-creation model that actively foster collaboration between Indigenous educators, linguists, and youth, is adapted to the their context, and complies with ethical guidelines. We outline an implementation plan with Bororo and Enawene Nawe communities to test our methods and, potentially, produce an AI-driven platform for Indigenous language education that is applicable across diverse sociolinguistic contexts in Brazil and beyond.
Effective flood risk management relies on accurate forecasting, yet the ”black box” nature of state-of-the-art Deep Learning models creates a barrier to trust and accountability in high-stakes public safety decisions. While existing Explainable AI (XAI) methods offer local attributions, they fail to provide the verifiable, operationally meaningful causal narratives required by disaster response authorities. To address this societal challenge, we propose Context-Aware Concept Distillation (CACD), a framework developed in collaboration with domain experts to distill opaque LSTMs into interpretable, hydrology-aware surrogate models. We introduce an unsupervised pipeline to discover a ”Hydrological Language” and a Residual Hypernetwork that dynamically modulates these concepts based on static basin characteristics. Evaluated on 5,203 basins globally, our model achieves high fidelity (Median NSE 0.70), significantly outperforming black-box baselines (e.g., Multi Layer Perceptrons) on unseen future data. By demonstrating that human-interpretable concepts are sufficient to reconstruct flood dynamics, this work balances AI accuracy with the transparency required for responsible environmental decision-making.
Codecfakes (CFs) are a type of speech deepfakes generated through Audio Language Models (ALMs), with Neural Audio Codecs (NACs) forming the core mechanism for speech encoding and generation. CFs exhibit distributional characteristics that differ from vocoder-based deepfakes, causing detectors trained on vocoder data to generalize poorly to CFs detection. Although this has led to the development of CF detection benchmarks, existing resources are largely confined to English—and to a limited extent Chinese—leaving South-East Asian (SEA) languages unexplored. To bridge this gap, we introduce SEA-CF, the first large-scale benchmark for CF detection spanning multiple SEA languages, diverse speaker profiles, and a wide range of NAC architectures. SEA-CF is constructed by synthesizing publicly available real speech corpora. Our experiments show that state-of-the-art (SOTA) CF detectors trained on English-centric datasets fail to generalize to SEA speech due to language-specific phonetic structures, tonal variations, and rich prosodic diversity. We further conduct a comprehensive zero-shot and fine-tuned evaluation of recent SOTA ALMs on SEA-CF. Fine-tuning the ALMs improves performance, however, these are very large being impractical for real-world application due to their scale, particularly in low-resource and latency-constrained settings. To address this limitation, we propose a novel small-ALM, GARUDA tailored for CF detection, which delivers strong performance while remaining lightweight. Extensive evaluations demonstrate that the proposed Small-ALM outperforms strong end-to-end and ALM-based baselines, establishing a new, practical direction for robust CF detection in SEA languages and beyond.
The rapid evolution of online scams, driven by transnational networks and mass-produced social engineering scenarios, has exposed the speed limitations of conventional detection, necessitating tighter inter-agency coordination. While LLMs show promise in scam identification, their role in accelerating integrated response frameworks remains underexplored. We propose Counter-Scam, a unified LLM-based multi-agent framework that orchestrates end-to-end response from initial detection to crime investigation. The framework first proposes safe data guidelines, emphasizing non-public scam data and secure dataset construction via scam-specific NER. Developed with insights from 37 stakeholders to reduce delays and improve analytical efficiency, the system integrates CSRA (multi-agent mitigation), CSRT (nine role-aligned NLP tasks), and CSRD (a corpus of 185,300 scam cases and 38,587 knowledge entries). Experiments show that fine-tuned sLLMs surpass commercial models with over 10% in all CSRT tasks and a 0.24 F1 improvement in scam-specific NER. This proves the framework's capability for enabling rapid, collaborative mitigation of online scam.
Improving Survey Participation in Low-Literacy Populations Through Value-Sensitive Conversational AI
Collecting reliable social data from low-literacy populations remains a persistent challenge, particularly when surveys involve sensitive topics and marginalized communities. Traditional paper-based and web-based survey modalities often suffer from high attrition and incomplete responses due to literacy barriers, social pressure, and interactional discomfort. In this paper, we present findings from an initial field evaluation comparing multiple survey modalities paper-based interviews, digital web-based surveys, conversational AI (convAI) surveys, and convAI enhanced with layered value-sensitive design conducted with low-literacy women across India. Using data from 315 participants, we show that convAI significantly improves survey completion rates relative to traditional modalities, with the highest completion and lowest drop-off observed when value-sensitive and culturally aligned conversational design elements are fully integrated. These results demonstrate the importance of human-centered and value-sensitive interaction design in enabling inclusive, ethical, and scalable data collection; motivating more `AI for social good' applications.