Humans display significant uncertainty when confronted with moral dilemmas, yet the extent of such uncertainty in machines and AI agents remains underexplored. Recent studies have confirmed the overly confident tendencies of machine-generated responses, particularly in large language models (LLMs). As these systems are increasingly embedded in ethical decision-making scenarios, it is important to understand their moral reasoning and the inherent uncertainties in building reliable AI systems. This work examines how uncertainty influences moral decisions in the classical trolley problem, analyzing responses from 32 open-source models and 9 distinct moral dimensions. We first find that variance in model confidence is greater across models than within moral dimensions, suggesting that moral uncertainty is predominantly shaped by model architecture and training method. To quantify uncertainty, we measure binary entropy as a linear combination of total entropy, conditional entropy, and mutual information. To examine its effects, we introduce stochasticity into models via ``dropout'' at inference time. Our findings show that our mechanism increases total entropy, mainly through a rise in mutual information, while conditional entropy remains largely unchanged. Moreover, this mechanism significantly improves human-LLM moral alignment, with correlations in mutual information and alignment score shifts. Our results highlight the potential to better align model-generated decisions and human preferences by deliberately modulating uncertainty and reducing LLMs' confidence in morally complex scenarios.

Satellite-based slum segmentation holds significant promise in generating global estimates of urban poverty. However, the morphological heterogeneity of informal settlements presents a major challenge, hindering the ability of models trained on specific regions to generalize effectively to unseen locations. To address this, we introduce a large-scale high-resolution dataset and propose GRAM (Generalized Region-Aware Mixture-of-Experts), a two-phase test-time adaptation framework that enables robust slum segmentation without requiring labeled data from target regions. We compile a million-scale satellite imagery dataset from 12 cities across four continents for source training. Using this dataset, the model employs a Mixture-of-Experts architecture to capture region-specific slum characteristics while learning universal features through a shared backbone. During adaptation, prediction consistency across experts filters out unreliable pseudo-labels, allowing the model to generalize effectively to previously unseen regions. GRAM outperforms state-of-the-art baselines in low-resource settings such as African cities, offering a scalable and label-efficient solution for global slum mapping and data-driven urban planning.

Open-weight large language models (LLMs) enable extensive customization but remain susceptible to post-release misuse via malicious fine-tuning.While existing defenses attempt to constrain parameter-space dynamics or mitigate harmful internal representations, malicious fine-tuning continues to erode these safeguards leaving the development of fundamental, persistent defenses for open-weight models an unresolved challenge.In this paper, we characterize a safety region for open-weight LLMs and propose Safety Guidance Trigger (SGT), a framework that preserves alignment by guiding finetuning toward the safety manifold.It has two stages: (1) optimizing a safety trigger to steer the base model outputs toward safe responses and ( 2) training the open-weight model to align its internal features with trigger-induced safety representations.We demonstrate that SGT substantially improves robustness against malicious fine-tuning, forcing adversaries to significantly increase data budgets to bypass safeguards.Our analysis further confirms that this approach anchors model representations within a safety region that remains resilient under adversarial attacks.

The increasing frequency and intensity of natural disasters call for rapid and accurate damage assessment. In response, disaster benchmark datasets from high-resolution satellite imagery have been constructed to develop methods for detecting damaged areas. However, these methods face significant challenges when applied to previously unseen regions due to the limited geographical and disaster-type diversity in the existing datasets. We introduce DAVI (Disaster Assessment with VIsion foundation model), a novel approach that addresses domain disparities and detects structural damage at the building level without requiring ground-truth labels for target regions. DAVI combines task-specific knowledge from a model trained on source regions with task-agnostic knowledge from an image segmentation model to generate pseudo labels indicating potential damage in target regions. It then utilizes a two-stage refinement process, which operate at both pixel and image levels, to accurately identify changes in disaster-affected areas. Our evaluation, including a case study on the 2023 Türkiye earthquake, demonstrates that our model achieves exceptional performance across diverse terrains (e.g., North America, Asia, and the Middle East) and disaster types (e.g., wildfires, hurricanes, and tsunamis). This confirms its robustness in disaster assessment without dependence on ground-truth labels and highlights its practical applicability.

International aid is a critical mechanism for promoting economic growth and well-being in developing nations, supporting progress toward the Sustainable Development Goals (SDGs). However, tracking aid contributions remains challenging due to labor-intensive data management, incomplete records, and the heterogeneous nature of aid data. Recognizing the urgency of this challenge, we partnered with government agencies to develop an AI model that complements manual classification and mitigates human bias in subjective interpretation. By integrating SDG-specific semantics and leveraging prior knowledge from language models, our approach enhances classification accuracy and accommodates the diversity of aid projects. When applied to a comprehensive dataset spanning multiple years, our model can reveal hidden trends in the temporal evolution of international development cooperation. Expert interviews further suggest how these insights can empower policymakers with data-driven decision-making tools, ultimately improving aid effectiveness and supporting progress toward SDGs.

The spread of fake news harms individuals and presents a critical social challenge that must be addressed. Although numerous algorithmic and insightful features have been developed to detect fake news, many of these features can be manipulated with style-conversion attacks, especially with the emergence of advanced language models, making it more difficult to differentiate from genuine news. This study proposes adversarial style augmentation, AdStyle, designed to train a fake news detector that remains robust against various style-conversion attacks. The primary mechanism involves the strategic use of LLMs to automatically generate a diverse and coherent array of style-conversion attack prompts, enhancing the generation of particularly challenging prompts for the detector. Experiments indicate that our augmentation strategy significantly improves robustness and detection performance when evaluated on fake news benchmark datasets.

Large language models have revolutionized text generation, offering significant benefits while also posing threats to society, such as copyright infringement and misinformation. To prevent harmful use, the task of detecting machine-generated content has become an important research topic, though it remains particularly challenging across diverse content domains. This paper presents DGRM, an innovative add-on module designed to improve the domain generalization capability of existing machine-generated text detectors. Our model consists of two training components. (1) Feature disentanglement separates a text's embedding into target-specific and common attributes, thereby enhancing semantic domain generalization across different content domains. (2) Feature regularization applies constraints to these attributes to extract additional target-relevant information and ensure detection consistency under syntactic perturbations—thus achieving syntactic domain generalization. Evaluation over multiple datasets demonstrates that incorporating our module substantially improves the detection of machine-generated text across semantically and syntactically diverse domains. We hope our work contributes to mitigating the harmful use of language models.

Federated learning is used to train a shared model in a decentralized way without clients sharing private data with each other. Federated learning systems are susceptible to poisoning attacks when malicious clients send false updates to the central server. Existing defense strategies are ineffective under non-IID data settings. This paper proposes a new defense strategy, FedCPA (Federated learning with Critical Parameter Analysis). Our attack-tolerant aggregation method is based on the observation that benign local models have similar sets of top-k and bottom-k critical parameters, whereas poisoned local models do not. Experiments with different attack scenarios on multiple datasets demonstrate that our model outperforms existing defense strategies in defending against poisoning attacks.

Federated learning enables learning from decentralized data sources without compromising privacy, which makes it a crucial technique. However, it is vulnerable to model poisoning attacks, where malicious clients interfere with the training process. Previous defense mechanisms have focused on the server-side by using careful model aggregation, but this may not be effective when the data is not identically distributed or when attackers can access the information of benign clients. In this paper, we propose a new defense mechanism that focuses on the client-side, called FedDefender, to help benign clients train robust local models and avoid the adverse impact of malicious model updates from attackers, even when a server-side defense cannot identify or remove adversaries. Our method consists of two main components: (1) attack-tolerant local meta update and (2) attack-tolerant global knowledge distillation. These components are used to find noise-resilient model parameters while accurately extracting knowledge from a potentially corrupted global model. Our client-side defense strategy has a flexible structure and can work in conjunction with any existing server-side strategies. Evaluations of real-world scenarios across multiple datasets show that the proposed method enhances the robustness of federated learning against model poisoning attacks.

We study the human-in-the-loop customs inspection scenario, where an AI-assisted algorithm supports customs officers by recommending a set of imported goods to be inspected. If the inspected items are fraudulent, the officers can levy extra duties. These logs are then used as additional training data for the next iterations. Choosing to inspect suspicious items first leads to an immediate gain in customs revenue, yet such inspections may not bring new insights for learning dynamic traffic patterns. On the other hand, inspecting uncertain items can help acquire new knowledge, which will be used as a supplementary training resource to update the selection systems. Based on multiyear customs datasets from three countries, we demonstrate that some degree of exploration is necessary to cope with domain shifts in the trade data. The results show that a hybrid strategy of selecting likely fraudulent and uncertain items will eventually outperform the exploitation-only strategy.

High-resolution daytime satellite imagery has become a promising source to study economic activities. These images display detailed terrain over large areas and allow zooming into smaller neighborhoods. Existing methods, however, have utilized images only in a single-level geographical unit. This research presents a deep learning model to predict economic indicators via aggregating traits observed from multiple levels of geographical units. The model first measures hyperlocal economy over small communities via ordinal regression. The next step extracts district-level features by summarizing interconnection among hyperlocal economies. In the final step, the model estimates economic indicators of districts via aggregating the hyperlocal and district information. Our new multi-level learning model substantially outperforms strong baselines in predicting key indicators such as population, purchasing power, and energy consumption. The model is also robust against data shortage; the trained features from one country can generalize to other countries when evaluated with data gathered from Malaysia, the Philippines, Thailand, and Vietnam. We discuss the multi-level model's implications for measuring inequality, which is the essential first step in policy and social science research on inequality and poverty.

Knowledge of the changing traffic is critical in risk management. Customs offices worldwide have traditionally relied on local resources to accumulate such knowledge and detect tax frauds. This naturally poses countries with weak infrastructure to become tax havens of potentially illicit trades. The current paper proposes DAS, a memory bank platform to facilitate knowledge sharing across multi-national customs administrations to support each other. We propose a domain adaptation method to share transferable knowledge of frauds as prototypes while safeguarding the local trade information. Data encompassing over 8 million import declarations have been used to test the feasibility of this new system, which shows that participating countries may benefit up to 2-11 times in fraud detection with the help of shared knowledge. We discuss implications for substantial tax revenue potential and strengthened policy against illicit trades.

Modern climate models offer simulation results that provide unprecedented details at the local level. However, even with powerful supercomputing facilities, their computational complexity and associated costs pose a limit on simulation resolution that is needed for agile planning of resource allocation, parameter calibration, and model reproduction. As regional information is vital for policymakers, data from coarse-grained resolution simulations undergo the process of "statistical downscaling" to generate higher-resolution projection at a local level. We present a new method for downscaling climate simulations called GINE (Geospatial INformation Encoded statistical downscaling). To preserve the characteristics of climate simulation data during this process, our model applies the latest computer vision techniques over topography-driven spatial and local-level information. The comprehensive evaluations on 2x, 4x, and 8x resolution factors show that our model substantially improves performance in terms of RMSE and the visual quality of downscaled data.

Unsupervised image clustering methods often introduce alternative objectives to indirectly train the model and are subject to faulty predictions and overconfident results. To overcome these challenges, the current research proposes an innovative model RUC that is inspired by robust learning. RUC’s novelty is at utilizing pseudo-labels of existing image clustering models as a noisy dataset that may include misclassified samples. Its retraining process can revise misaligned knowledge and alleviate the overconfidence problem in predictions. The model’s flexible structure makes it possible to be used as an add-on module to other clustering methods and helps them achieve better performance on multiple datasets. Extensive experiments show that the proposed model can adjust the model confidence with better calibration and gain additional robustness against adversarial noise.

Satellite imagery has long been an attractive data source providing a wealth of information regarding human-inhabited areas. While high-resolution satellite images are rapidly becoming available, limited studies have focused on how to extract meaningful information regarding human habitation patterns and economic scales from such data. We present READ, a new approach for obtaining essential spatial representation for any given district from high-resolution satellite imagery based on deep neural networks. Our method combines transfer learning and embedded statistics to efficiently learn the critical spatial characteristics of arbitrary size areas and represent such characteristics in a fixed-length vector with minimal information loss. Even with a small set of labels, READ can distinguish subtle differences between rural and urban areas and infer the degree of urbanization. An extensive evaluation demonstrates that the model outperforms state-of-the-art models in predicting economic scales, such as the population density in South Korea (R2=0.9617), and shows a high use potential in developing countries where district-level economic scales are unknown.

Self-supervised representation learning aims to generate effective representations for data instances without the need for manual labels, also known as unsupervised embedding learning, which has been a critical challenge in many existing semi-supervised and supervised learning tasks. This paper proposes a new self-supervised learning approach, called Super-AND, which extends the memory-based pretraining method AND model [13]. Super-AND has its unique set of losses that combines data augmentation in neighborhood discovery for more accurate anchor selection in embedding learning and further presents an adversarial training manner to learn more confident embeddings under the unsupervised setting. Experimental results exhibit that Super-AND outperforms all existing state-of-the-art self-supervised representation learning approaches and achieves an accuracy of 89.2% on the image classification task for CIFAR-10.

Reliable and timely measurements of economic activities are fundamental for understanding economic development and designing government policies. However, many developing countries still lack reliable data. In this paper, we introduce a novel approach for measuring economic development from high-resolution satellite images in the absence of ground truth statistics. Our method consists of three steps. First, we run a clustering algorithm on satellite images that distinguishes artifacts from nature (siCluster). Second, we generate a partial order graph of the identified clusters based on the level of economic development, either by human guidance or by low-resolution statistics (siPog). Third, we use a CNN-based sorter that assigns differentiable scores to each satellite grid based on the relative ranks of clusters (siScore). The novelty of our method is that we break down a computationally hard problem into sub-tasks, which involves a human-in-the-loop solution. With the combination of unsupervised learning and the partial orders of dozens of urban vs. rural clusters, our method can estimate the economic development scores of over 10,000 satellite grids consistently with other baseline development proxies (Spearman correlation of 0.851). This efficient method is interpretable and robust; we demonstrate how to apply our method to both developed (e.g., South Korea) and developing economies (e.g., Vietnam and Malawi).