Search: “safety”

Why dynamically routing multi-timescale advantages in PPO causes policy collapse (and a simple decoupled fix) [R]

This Reddit post discusses a known instability in PPO when advantage estimates operating across different temporal scales (e.g., short-horizon and long-horizon returns) are dynamically routed or mixed

safetyr-machinelearning

16 Apr 2026

Human

Content type

All Blog X Post Paper YouTube Reddit GitHub

Clear filters

Safety

Why MLLMs Struggle to Determine Object Orientations

DGX agent

arXiv:2604.13321v1 Announce Type: new Abstract: Multimodal Large Language Models (MLLMs) struggle with tasks that require reasoning about 2D object orientation in images, as documented in prior work.

safetyarxiv-cs-cv

16 Apr 2026

Why Multimodal In-Context Learning Lags Behind? Unveiling the Inner Mechanisms and Bottlenecks

arXiv:2604.13403v1 Announce Type: new Abstract: In-context learning (ICL) enables models to adapt to new tasks via inference-time demonstrations. Despite its success in large language models, the exte

16 Apr 2026

5/5 The takeaway: If your agent relies on an LLM judge for selection accuracy, measuring code quality isn’t enough; you need a measure of th…

5/5 The takeaway: If your agent relies on an LLM judge for selection accuracy, measuring code quality isn’t enough; you need a measure of the model's inductive bias toward the 'fingerprint' of a gold

safetyai21-labs--x

A Dataset and Evaluation for Complex 4D Markerless Human Motion Capture

arXiv:2604.12765v1 Announce Type: new Abstract: Marker-based motion capture (MoCap) systems have long been the gold standard for accurate 4D human modeling, yet their reliance on specialized hardware

A document is worth a structured record: Principled inductive bias design for document recognition

arXiv:2507.08458v2 Announce Type: replace-cross Abstract: Many document types use intrinsic, convention-driven structures that serve to encode precise and structured information, such as the conventio

A Survey of Multimodal Mathematical Reasoning: From Perception, Alignment to Reasoning

arXiv:2603.08291v3 Announce Type: replace Abstract: Multimodal Mathematical Reasoning (MMR) has recently attracted increasing attention for its capability to solve mathematical problems involving both

A TTP analysis finds dozens of nudify apps in Apple and Google app stores via search, despite company policy prohibiting them; the apps earned $122M+ in revenue (Bloomberg)

Bloomberg: A TTP analysis finds dozens of nudify apps in Apple and Google app stores via search, despite company policy prohibiting them; the apps earned $122M+ in revenue — Apple Inc. and Google have

safetytechmeme

A Two-Timescale Primal-Dual Framework for Reinforcement Learning via Online Dual Variable Guidance

arXiv:2505.04494v3 Announce Type: replace-cross Abstract: We study reinforcement learning by combining recent advances in regularized linear programming formulations with the classical theory of stoch

safetyarxiv-cs-lg

AAPO: Enhancing the Reasoning Capabilities of LLMs with Advantage Margin

arXiv:2505.14264v3 Announce Type: replace-cross Abstract: Reinforcement learning (RL) has emerged as an effective approach for enhancing the reasoning capabilities of large language models (LLMs), esp

Advancing Multi-Agent RAG Systems with Minimalist Reinforcement Learning

arXiv:2505.17086v4 Announce Type: replace Abstract: Large Language Models (LLMs) equipped with modern Retrieval-Augmented Generation (RAG) systems often employ multi-turn interaction pipelines to inte

ART-VITON: Measurement-Guided Latent Diffusion for Artifact-Free Virtual Try-On

arXiv:2509.25749v2 Announce Type: cross Abstract: Virtual try-on (VITON) aims to generate realistic images of a person wearing a target garment, requiring precise garment alignment in try-on regions a

BayMOTH: Bayesian optiMizatiOn with meTa-lookahead -- a simple approacH

arXiv:2604.12005v1 Announce Type: cross Abstract: Bayesian optimization (BO) has for sequential optimization of expensive black-box functions demonstrated practicality and effectiveness in many real-w

Beyond Factual Grounding: The Case for Opinion-Aware Retrieval-Augmented Generation

arXiv:2604.12138v1 Announce Type: new Abstract: RAG systems have transformed how LLMs access external knowledge, but we find that current implementations exhibit a bias toward factual, objective conte

Beyond Transcription: Unified Audio Schema for Perception-Aware AudioLLMs

arXiv:2604.12506v1 Announce Type: new Abstract: Recent Audio Large Language Models (AudioLLMs) exhibit a striking performance inversion: while excelling at complex reasoning tasks, they consistently u

Black-Box Optimization From Small Offline Datasets via Meta Learning with Synthetic Tasks

arXiv:2604.12325v1 Announce Type: cross Abstract: We consider the problem of offline black-box optimization, where the goal is to discover optimal designs (e.g., molecules or materials) from past expe

BRAIN: Bias-Mitigation Continual Learning Approach to Vision-Brain Understanding

arXiv:2508.18187v2 Announce Type: replace-cross Abstract: Memory decay makes it harder for the human brain to recognize visual objects and retain details. Consequently, recorded brain signals become w

Brain-DiT: A Universal Multi-state fMRI Foundation Model with Metadata-Conditioned Pretraining

arXiv:2604.12683v1 Announce Type: new Abstract: Current fMRI foundation models primarily rely on a limited range of brain states and mismatched pretraining tasks, restricting their ability to learn ge

Bridging the Micro--Macro Gap: Frequency-Aware Semantic Alignment for Image Manipulation Localization

arXiv:2604.12341v1 Announce Type: new Abstract: As generative image editing advances, image manipulation localization (IML) must handle both traditional manipulations with conspicuous forensic artifac

Calibration-Aware Policy Optimization for Reasoning LLMs

arXiv:2604.12632v1 Announce Type: cross Abstract: Group Relative Policy Optimization (GRPO) enhances LLM reasoning but often induces overconfidence, where incorrect responses yield lower perplexity th

CamReasoner: Reinforcing Camera Movement Understanding via Structured Spatial Reasoning

arXiv:2602.00181v3 Announce Type: replace-cross Abstract: Understanding camera dynamics is a fundamental pillar of video spatial intelligence. However, existing multimodal models predominantly treat t

Causal Diffusion Models for Counterfactual Outcome Distributions in Longitudinal Data

arXiv:2604.12992v1 Announce Type: cross Abstract: Predicting counterfactual outcomes in longitudinal data, where sequential treatment decisions heavily depend on evolving patient states, is critical y

safetyarxiv-cs-lg

Challenging Vision-Language Models with Physically Deployable Multimodal Semantic Lighting Attacks

arXiv:2604.12833v1 Announce Type: new Abstract: Vision-Language Models (VLMs) have shown remarkable performance, yet their security remains insufficiently understood. Existing adversarial studies focu

CIA: Inferring the Communication Topology from LLM-based Multi-Agent Systems

arXiv:2604.12461v1 Announce Type: new Abstract: LLM-based Multi-Agent Systems (MAS) have demonstrated remarkable capabilities in solving complex tasks. Central to MAS is the communication topology whi

CLEAR: Cross-Lingual Enhancement in Alignment via Reverse-training

arXiv:2604.05821v2 Announce Type: replace Abstract: Existing multilingual embedding models often encounter challenges in cross-lingual scenarios due to imbalanced linguistic resources and less conside

Combating Pattern and Content Bias: Adversarial Feature Learning for Generalized AI-Generated Image Detection

arXiv:2604.12353v1 Announce Type: new Abstract: In recent years, the rapid development of generative artificial intelligence technology has significantly lowered the barrier to creating high-quality f

Contextual Multi-Task Reinforcement Learning for Autonomous Reef Monitoring

arXiv:2604.12645v1 Announce Type: cross Abstract: Although autonomous underwater vehicles promise the capability of marine ecosystem monitoring, their deployment is fundamentally limited by the diffic

Continuous Knowledge Metabolism: Generating Scientific Hypotheses from Evolving Literature

arXiv:2604.12243v1 Announce Type: cross Abstract: Scientific hypothesis generation requires tracking how knowledge evolves, not just what is currently known. We introduce Continuous Knowledge Metaboli

CoSyncDiT: Cognitive Synchronous Diffusion Transformer for Movie Dubbing

arXiv:2604.12292v1 Announce Type: cross Abstract: Movie dubbing aims to synthesize speech that preserves the vocal identity of a reference audio while synchronizing with the lip movements in a target

Cross-Cultural Simulation of Citizen Emotional Responses to Bureaucratic Red Tape Using LLM Agents

arXiv:2604.12545v1 Announce Type: new Abstract: Improving policymaking is a central concern in public administration. Prior human subject studies reveal substantial cross-cultural differences in citiz

Cross-Modal Knowledge Distillation for PET-Free Amyloid-Beta Detection from MRI

arXiv:2604.12574v1 Announce Type: new Abstract: Detecting amyloid-eta (Aeta) positivity is crucial for early diagnosis of Alzheimer's disease but typically requires PET imaging, which is costly, invas

Cycle-Consistent Search: Question Reconstructability as a Proxy Reward for Search Agent Training

arXiv:2604.12967v1 Announce Type: new Abstract: Reinforcement Learning (RL) has shown strong potential for optimizing search agents in complex information retrieval tasks. However, existing approaches

DBGL: Decay-aware Bipartite Graph Learning for Irregular Medical Time Series Classification

arXiv:2604.11842v1 Announce Type: cross Abstract: Irregular Medical Time Series play a critical role in the clinical domain to better understand the patient's condition. However, inherent irregularity

Designing Reliable LLM-Assisted Rubric Scoring for Constructed Responses: Evidence from Physics Exams

arXiv:2604.12227v1 Announce Type: new Abstract: Student responses in STEM assessments are often handwritten and combine symbolic expressions, calculations, and diagrams, creating substantial variation

DocSeeker: Structured Visual Reasoning with Evidence Grounding for Long Document Understanding

arXiv:2604.12812v1 Announce Type: new Abstract: Existing Multimodal Large Language Models (MLLMs) suffer from significant performance degradation on the long document understanding task as document le

Does RLVR Extend Reasoning Boundaries? Investigating Capability Expansion in Vision-Language Models

arXiv:2511.00710v4 Announce Type: replace Abstract: Recent studies posit that Reinforcement Learning with Verifiable Rewards (RLVR) primarily amplifies behaviors inherent to the pre-training distribut

DyBBT: Dynamic Balance via Bandit-inspired Targeting for Dialog Policy with Cognitive Dual-Systems

arXiv:2509.19695v3 Announce Type: replace-cross Abstract: Task oriented dialog systems often rely on static exploration strategies that do not adapt to dynamic dialog contexts, leading to inefficient

Dynamic Multi-Robot Task Allocation under Uncertainty and Communication Constraints: A Game-Theoretic Approach

arXiv:2604.11954v1 Announce Type: cross Abstract: We study dynamic multi-robot task allocation under uncertain task completion, time-window constraints, and incomplete information. Tasks arrive online

safetyarxiv-cs-ro

E2E-Fly: An Integrated Training-to-Deployment System for End-to-End Quadrotor Autonomy

arXiv:2604.12916v1 Announce Type: new Abstract: Training and transferring learning-based policies for quadrotors from simulation to reality remains challenging due to inefficient visual rendering, phy

safetyarxiv-cs-ro

Enhancing Agentic Textual Graph Retrieval with Synthetic Stepwise Supervision

arXiv:2510.03323v2 Announce Type: replace Abstract: Integrating textual graphs into Large Language Models (LLMs) is promising for complex graph-based QA. However, a key bottleneck is retrieving inform

Euler-inspired Decoupling Neural Operator for Efficient Pansharpening

arXiv:2604.12463v1 Announce Type: cross Abstract: Pansharpening aims to synthesize high-resolution multispectral (HR-MS) images by fusing the spatial textures of panchromatic (PAN) images with the spe

Evaluating the Limitations of Protein Sequence Representations for Parkinson's Disease Classification

arXiv:2604.11852v1 Announce Type: cross Abstract: The identification of reliable molecular biomarkers for Parkinson's disease remains challenging due to its multifactorial nature. Although protein seq

Evolution-Inspired Sample Competition for Deep Neural Network Optimization

arXiv:2604.12568v1 Announce Type: new Abstract: Conventional deep network training generally optimizes all samples under a largely uniform learning paradigm, without explicitly modeling the heterogene

EvoSpark: Endogenous Interactive Agent Societies for Unified Long-Horizon Narrative Evolution

arXiv:2604.12776v1 Announce Type: new Abstract: Realizing endogenous narrative evolution in LLM-based multi-agent systems is hindered by the inherent stochasticity of generative emergence. In particul

From Myopic Selection to Long-Horizon Awareness: Sequential LLM Routing for Multi-Turn Dialogue

arXiv:2604.12385v1 Announce Type: new Abstract: Multi-turn dialogue is the predominant form of interaction with large language models (LLMs). While LLM routing is effective in single-turn settings, ex

Generative AI in nutshell: 'If the technology is as formidable as the [tech CEO's] claim, then they could be leading us toward existential d…

Generative AI in nutshell: 'If the technology is as formidable as the [tech CEO's] claim, then they could be leading us toward existential disaster; if the technology proves less transformative, and t

safetygary-marcus--x

GeoAlign: Geometric Feature Realignment for MLLM Spatial Reasoning

arXiv:2604.12630v1 Announce Type: cross Abstract: Multimodal large language models (MLLMs) have exhibited remarkable performance in various visual tasks, yet still struggle with spatial reasoning. Rec

Gradient boundaries through confidence intervals for forced alignment estimates using model ensembles

arXiv:2506.01256v4 Announce Type: replace-cross Abstract: Forced alignment is a common tool to align audio with orthographic and phonetic transcriptions. Most forced alignment tools provide only point