Alignment, interpretability, red-teaming, bias and privacy: the research on what these systems do when they misbehave.
arXiv:2608.24531v1 Announce Type: cross Abstract: Probabilistic full-field reconstruction provides uncertainty-aware response evidence for structural reliability assessment, yet inference from sparse...
The paper examines whether benign overfitting—where highly overparameterized models still predict well—occurs in equity return prediction. It finds a double‑descent risk curve for ridgeless models and shows that while ridge regularization slightly improves performance, the advantage vanishes at high parameter‑to‑observation ratios. Ultimately, both models fail to beat a simple historical average, indicating that standard equity predictors lack genuine forecasting power even with flexible machine learning methods.
The paper introduces Evidence-First Reflection (EFR), a two-stage reflector for desktop GUI agents that separates visual difference extraction from outcome verification. EFR uses Set-of-Marks annotations to locate action sites and candidate changed regions, filters action-relevant changes, and then makes a final judgment based on cleaned evidence. Experiments on OSWorld-Verified and WindowsAgentArena show that EFR improves reflector accuracy by 7.11% and increases end-to-end task success by roughly 5–6%.
The paper evaluates the use of large language models (LLMs) as judges for assessing conversational voice agents, comparing human judgments with GPT‑4.1 and GPT‑5 across telecom and retail interactions. It examines agreement, metric‑level correlations, and consistency across three evaluation configurations (p0, p1, p2) to determine how reliably LLMs can judge conversational quality and safety. The study finds that LLM‑based evaluation can be effective but its reliability varies by metric and configuration, suggesting a hybrid approach where LLMs handle scalable assessment while humans focus on metrics requiring contextual interpretation.
The paper introduces MS‑WDRO, a multi‑source Wasserstein distributionally robust optimization framework for reconstructing complex network topologies from scarce target‑domain data and abundant heterogeneous source data. It fuses sources via a weighted Wasserstein barycenter, builds an ambiguity set around it, and solves a regularized Laplacian estimator using a provably convergent ADMM scheme. The authors provide finite‑sample guarantees, demonstrate that naive aggregation is suboptimal, and show through experiments on synthetic data and the ABIDE I neuroimaging dataset that MS‑WDRO outperforms seven baselines in graph recovery, sample efficiency, and diagnostic utility, especially when target samples are limited.
The paper introduces a two‑layer detector to prevent ‘slopsquatting’—the risk of local coding LLMs fabricating Python package names that adversaries can pre‑register on PyPI. The first layer checks PyPI for existence, while the second uses a Random Forest classifier on ten name‑and‑metadata features; an import reconciler resolves naming mismatches. Embedded in a LangGraph state machine, the system retries at escalating temperatures and falls back to stronger models, achieving hallucination‑free code in 76% of 300 curated prompts and recovering additional runs through intra‑ and cross‑model retries.
arXiv:2608.24492v1 Announce Type: cross Abstract: Uncertainty quantification (UQ) methods are widely used for hallucination detection in large language models (LLMs) in closed-book settings where gro...
arXiv:2608.24885v1 Announce Type: cross Abstract: Action-conditioned world models are increasingly used as learned simulators for policy evaluation and improvement, yet their effectiveness rests on a...
arXiv:2608.24631v1 Announce Type: cross Abstract: Similarity in many decision systems is governed not by distance alone but by interactions among variables. In fraud and anomaly detection, small loca...
The paper introduces RobustTests, a framework that improves reinforcement learning for code generation by synthesizing test cases from faulty code and refining rewards with a dense, stepwise function. It uses validator agents and behavioral clustering to filter out invalid or redundant tests, and incorporates pass‑rate‑based rewards to counter hallucination noise. Experiments on CodeContests and LiveCodeBench show that fine‑tuning Qwen3‑32B with RobustTests yields a 3% absolute performance gain over baseline methods.
The paper audits a developer‑accessible on‑device language model, revealing that it can confidently produce incorrect answers while refusing benign prompts, a phenomenon termed task‑asymmetric miscalibration. The model’s confident outputs are surface‑indistinguishable, with classifiers based on user‑visible features failing to separate correct from wrong responses. The authors propose a model‑agnostic audit protocol, a surface‑indistinguishability test, and a black‑box consistency wrapper that improves reliability without requiring model access.
PHMForge is an evaluation environment that tests large language model agents on industrial prognostics tasks using the Model Context Protocol (MCP). It provides 99 SME-authored scenarios across eight asset classes, 39 MCP-native tools wrapping published PHM algorithms, and deterministic evaluators that separate protocol fluency, reasoning, instrumentation, and tool use. The benchmark shows that the best agent configuration achieves 80.8% pass@1, with orchestration and tool sequencing errors as the main failure modes, and demonstrates the limitations of static retrieval for prognostic computation.
arXiv:2511.02605v3 Announce Type: replace Abstract: Shielding is widely used to enforce safety in reinforcement learning (RL), ensuring that an agent's actions remain compliant with formal specificat...
arXiv:2608.24282v1 Announce Type: new Abstract: Adaptive LiDAR scanning concentrates a limited sensing budget on regions of interest predicted from past object tracks, lowering data volume in autonom...
The paper introduces Restoring without Forgetting (RwF), a continual learning framework for image restoration that handles multiple degradations sequentially without accessing prior data. RwF trains a lightweight adapter for each new degradation, uses an unsupervised routing mechanism to select the correct restoration path, and achieves significant PSNR gains over fine‑tuning on five benchmark degradation domains. The method also demonstrates strong transfer performance on eleven real‑degradation datasets with high routing accuracy.
The paper introduces CloudEmu, an automated system that builds cloud emulators by synthesizing code from cloud documentation using a neurosymbolic approach. It combines large language models for understanding documentation with symbolic abstractions to reduce hallucinations and ensure precision, and it employs the real cloud as an oracle for testing and repair. Evaluation shows CloudEmu achieves higher coverage and accuracy than the manually developed LocalStack for AWS and GCP services.
arXiv:2608.24807v1 Announce Type: cross Abstract: Model cards are structured documents that summarize key information about machine learning models to improve transparency, usability, and accountabil...
The paper introduces Variance‑Guided Spatial Attention Fusion (VG‑SAF), a method for robust end‑to‑end driving that fuses camera and LiDAR data while handling asymmetric sensor degradation. VG‑SAF uses a physically grounded augmentor to generate dense reliability masks, modality‑specific experts to predict per‑pixel reliability scales, and a hybrid attention mechanism that gates unreliable cells and balances modalities. The approach also includes a Laplace uncertainty head to signal severe or combined sensor failures, and demonstrates improved closed‑loop robustness on the CARLA Longest6 benchmark across various degradation scenarios.
arXiv:2606.00930v2 Announce Type: replace-cross Abstract: Mechanistic interpretability routinely reads a probe and labels its top-activating units as the circuit executing the computation. We test th...
ESQ‑Bench is an Oracle‑first NL2SQL benchmark that introduces systematic complexity tiers and silent‑divergence evaluation across three enterprise schema levels. It provides six populated schemas (465 tables, 164,682 rows) on Oracle, PostgreSQL, MySQL, and SQL Server, along with 550 gold‑validated question‑query pairs and a four‑metric evaluation harness. The benchmark reveals that state‑of‑the‑art models such as GPT‑4o and Claude Sonnet 4.6 experience significant drops in execution accuracy and high silent‑divergence rates as schema complexity increases, highlighting a gap between closed‑API models and open‑weight baselines on enterprise Oracle schemas.