On-the-fly reasoning often requires adaptation to novel problems under limited data and distribution shift. This work introduces CausalARC: an experimental testbed for AI reasoning in low-data and out-of-distribution regimes, modeled after the Abstraction and Reasoning Corpus (ARC). Each CausalARC reasoning task is sampled from a fully specified causal world model, formally expressed as a structural causal model. Principled data augmentations provide observational, interventional, and counterfactual feedback about the world model in the form of few-shot, in-context learning demonstrations. As a proof-of-concept, we illustrate the use of CausalARC for four language model evaluation settings: (1) abstract reasoning with test-time training, (2) counterfactual reasoning with in-context learning, (3) program synthesis, and (4) causal discovery with logical reasoning. Within- and between-model performance varied heavily across tasks, indicating room for significant improvement in language model reasoning.

Self-supervised monocular depth estimation is a significant task for low-cost and efficient 3D scene perception and measurement in endoscopy. However, the variety of illumination conditions and scene features is still the primary challenges for depth estimation in endoscopic scenes. In this work, a novel self-supervised framework is proposed for monocular depth estimation in diverse endoscopy. Firstly, considering the diverse features in endoscopic scenes with different tissues, a novel block-wise mixture of dynamic low-rank experts is proposed to efficiently finetune the foundation model for endoscopic depth estimation. In the proposed module, based on the input feature, different experts with a small amount of trainable parameters are adaptively selected for weighted inference, from low-rank experts which are allocated based on the generalization of each block. Moreover, a novel self-supervised training framework is proposed to jointly cope with brightness inconsistency and reflectance interference. The proposed method outperforms state-of-the-art works on SCARED dataset and SimCol dataset. Furthermore, the proposed network also achieves the best generalization based on zero-shot depth estimation on C3VD, Hamlyn and SERV-CT dataset. The outstanding performance of our model is further demonstrated with 3D reconstruction and ego-motion estimation. The proposed method could contribute to accurate endoscopy for minimally invasive measurement and surgery. The evaluation codes will be released upon acceptance, while the demo videos can be found on: https://endo-gmde.netlify.app/.

We study the problem of opinion highlights generation from large volumes of user reviews, often exceeding thousands per entity, where existing methods either fail to scale or produce generic, one-size-fits-all summaries that overlook personalized needs. To tackle this, we introduce OpinioRAG, a scalable, training-free framework that combines RAG-based evidence retrieval with LLMs to efficiently produce tailored summaries. Additionally, we propose novel reference-free verification metrics designed for sentiment-rich domains, where accurately capturing opinions and sentiment alignment is essential. These metrics offer a fine-grained, context-sensitive assessment of factual consistency. To facilitate evaluation, we contribute the first large-scale dataset of long-form user reviews, comprising entities with over a thousand reviews each, paired with unbiased expert summaries and manually annotated queries. Through extensive experiments, we identify key challenges, provide actionable insights into improving systems, pave the way for future research, and position OpinioRAG as a robust framework for generating accurate, relevant, and structured summaries at scale.

Khiops is an open source machine learning tool designed for mining large multi-table databases. Khiops is based on a unique Bayesian approach that has attracted academic interest with more than 20 publications on topics such as variable selection, classification, decision trees and co-clustering. It provides a predictive measure of variable importance using discretisation models for numerical data and value clustering for categorical data. The proposed classification/regression model is a naive Bayesian classifier incorporating variable selection and weight learning. In the case of multi-table databases, it provides propositionalisation by automatically constructing aggregates. Khiops is adapted to the analysis of large databases with millions of individuals, tens of thousands of variables and hundreds of millions of records in secondary tables. It is available on many environments, both from a Python library and via a user interface.

GPUs running deep learning (DL) workloads are frequently underutilized. Collocating multiple DL training tasks on the same GPU can improve utilization but introduces two key risks: (1) out-of-memory (OOM) crashes for newly scheduled tasks, and (2) severe performance interference among co-running tasks, which can negate any throughput gains. These issues reduce system robustness, quality of service, and energy efficiency. We present CARMA, a task-level, collocation-aware resource management system for the server-scale. CARMA addresses collocation challenges via (1) fine-grained monitoring and bookkeeping of GPUs and a collocation risk analysis that filters out the high-risk GPUs; (2) task placement policies that cap GPU utilization to avoid OOMs and limit interference; (3) integration of GPU memory need estimators for DL tasks to minimize OOMs during collocation; and (4) a lightweight recovery method that relaunches jobs crashed due to OOMs. Our evaluation on a DL training workload derived from real-world traces shows that CARMA uses GPUs more efficiently by making more informed collocation decisions: for the best-performing collocation policy, CARMA increases GPU streaming multiprocessor (SM) utilization by 54%, the parallelism achieved per SM by 61%, and memory use by 62%. This results in a $\sim$35% and $\sim$15% reduction in the end-to-end execution time (makespan) and GPU energy consumption, respectively, for this workload.

Court judgments reveal how legal rules have been interpreted and applied to facts, providing a foundation for understanding structured legal reasoning. However, existing automated approaches for capturing legal reasoning, including large language models, often fail to identify the relevant legal context, do not accurately trace how facts relate to legal norms, and may misrepresent the layered structure of judicial reasoning. These limitations hinder the ability to capture how courts apply the law to facts in practice. In this paper, we address these challenges by constructing a legal knowledge graph from 648 Japanese administrative court decisions. Our method extracts components of legal reasoning using prompt-based large language models, normalizes references to legal provisions, and links facts, norms, and legal applications through an ontology of legal inference. The resulting graph captures the full structure of legal reasoning as it appears in real court decisions, making implicit reasoning explicit and machine-readable. We evaluate our system using expert annotated data, and find that it achieves more accurate retrieval of relevant legal provisions from facts than large language model baselines and retrieval-augmented methods.

The Lottery Ticket Hypothesis (LTH) suggests that within large neural networks, there exist sparse, trainable "winning tickets" capable of matching the performance of the full model, but identifying them through Iterative Magnitude Pruning (IMP) is computationally expensive. Recent work introduced COLT, an accelerator that discovers a "consensus" subnetwork by intersecting masks from models trained on disjoint data partitions; however, this approach discards all non-overlapping weights, assuming they are unimportant. This paper challenges that assumption and proposes the Disentangled Lottery Ticket (DiLT) Hypothesis, which posits that the intersection mask represents a universal, task-agnostic "core" subnetwork, while the non-overlapping difference masks capture specialized, task-specific "specialist" subnetworks. A framework is developed to identify and analyze these components using the Gromov-Wasserstein (GW) distance to quantify functional similarity between layer representations and reveal modular structures through spectral clustering. Experiments on ImageNet and fine-grained datasets such as Stanford Cars, using ResNet and Vision Transformer architectures, show that the "core" ticket provides superior transfer learning performance, the "specialist" tickets retain domain-specific features enabling modular assembly, and the full re-assembled "union" ticket outperforms COLT - demonstrating that non-consensus weights play a critical functional role. This work reframes pruning as a process for discovering modular, disentangled subnetworks rather than merely compressing models.

Large Language Models (LLMs) remain vulnerable to jailbreak attacks, which attempt to elicit harmful responses from LLMs. The evolving nature and diversity of these attacks pose many challenges for defense systems, including (1) adaptation to counter emerging attack strategies without costly retraining, and (2) control of the trade-off between safety and utility. To address these challenges, we propose Retrieval-Augmented Defense (RAD), a novel framework for jailbreak detection that incorporates a database of known attack examples into Retrieval-Augmented Generation, which is used to infer the underlying, malicious user query and jailbreak strategy used to attack the system. RAD enables training-free updates for newly discovered jailbreak strategies and provides a mechanism to balance safety and utility. Experiments on StrongREJECT show that RAD substantially reduces the effectiveness of strong jailbreak attacks such as PAP and PAIR while maintaining low rejection rates for benign queries. We propose a novel evaluation scheme and show that RAD achieves a robust safety-utility trade-off across a range of operating points in a controllable manner.

Large language models (LLMs) have demonstrated impressive capabilities across diverse tasks, but they remain susceptible to hallucinations--generating content that appears plausible but contains factual inaccuracies. We present Finch-Zk, a black-box framework that leverages fine-grained cross-model consistency to detect and mitigate hallucinations in LLM outputs without requiring external knowledge sources. Finch-Zk introduces two key innovations: 1) a cross-model consistency checking strategy that reveals fine-grained inaccuracies by comparing responses generated by diverse models from semantically-equivalent prompts, and 2) a targeted mitigation technique that applies precise corrections to problematic segments while preserving accurate content. Experiments on the FELM dataset show Finch-Zk improves hallucination detection F1 scores by 6-39\% compared to existing approaches. For mitigation, Finch-Zk achieves up to 9 absolute percentage points improvement in answer accuracy on the GPQA-diamond dataset when applied to state-of-the-art models like Llama 4 Maverick and Claude 4 Sonnet. Extensive evaluation on multiple datasets demonstrates that Finch-Zk provides a practical, deployment-ready safeguard for enhancing factual reliability in production LLM systems.

Earth observation analytics have the potential to serve many time-sensitive applications. However, due to limited bandwidth and duration of ground-satellite connections, it takes hours or even days to download and analyze data from existing Earth observation satellites, making real-time demands like timely disaster response impossible. Toward real-time analytics, we introduce OrbitChain, a collaborative analytics framework that orchestrates computational resources across multiple satellites in an Earth observation constellation. OrbitChain decomposes analytics applications into microservices and allocates computational resources for time-constrained analysis. A traffic routing algorithm is devised to minimize the inter-satellite communication overhead. OrbitChain adopts a pipeline workflow that completes Earth observation tasks in real-time, facilitates time-sensitive applications and inter-constellation collaborations such as tip-and-cue. To evaluate OrbitChain, we implement a hardware-in-the-loop orbital computing testbed. Experiments show that our system can complete up to 60% analytics workload than existing Earth observation analytics framework while reducing the communication overhead by up to 72%.