AI News — Latest Developments & Breakthroughs

ArXiv cs.AI 🧠 Large Language Models 📄 Paper ⚡ AI Lesson 11h ago

Good Scores, Bad Data: A Metric for Multimodal Coherence

arXiv:2603.25924v1 Announce Type: cross Abstract: Multimodal AI systems are evaluated by downstream task accuracy, but high accuracy does not mean the underlyin

ArXiv cs.AI 🧠 Large Language Models 📄 Paper ⚡ AI Lesson 11h ago

DiReCT: Disentangled Regularization of Contrastive Trajectories for Physics-Refined Video Generation

arXiv:2603.25931v1 Announce Type: cross Abstract: Flow-matching video generators produce temporally coherent, high-fidelity outputs yet routinely violate elemen

ArXiv cs.AI 👁️ Computer Vision 📄 Paper ⚡ AI Lesson 11h ago

DenseSwinV2: Channel Attentive Dual Branch CNN Transformer Learning for Cassava Leaf Disease Classification

arXiv:2603.25935v1 Announce Type: cross Abstract: This work presents a new Hybrid Dense SwinV2, a two-branch framework that jointly leverages densely connected

ArXiv cs.AI 🧠 Large Language Models 📄 Paper ⚡ AI Lesson 11h ago

Reinforcing Structured Chain-of-Thought for Video Understanding

arXiv:2603.25942v1 Announce Type: cross Abstract: Multi-modal Large Language Models (MLLMs) show promise in video understanding. However, their reasoning often

ArXiv cs.AI 📄 Paper 11h ago

Can Small Models Reason About Legal Documents? A Comparative Study

arXiv:2603.25944v1 Announce Type: cross Abstract: Large language models show promise for legal applications, but deploying frontier models raises concerns about

ArXiv cs.AI 👁️ Computer Vision 📄 Paper ⚡ AI Lesson 11h ago

Collision-Aware Vision-Language Learning for End-to-End Driving with Multimodal Infraction Datasets

arXiv:2603.25946v1 Announce Type: cross Abstract: High infraction rates remain the primary bottleneck for end-to-end (E2E) autonomous driving, as evidenced by t

ArXiv cs.AI 🧠 Large Language Models 📄 Paper ⚡ AI Lesson 11h ago

When Chain-of-Thought Backfires: Evaluating Prompt Sensitivity in Medical Language Models

arXiv:2603.25960v1 Announce Type: cross Abstract: Large Language Models (LLMs) are increasingly deployed in medical settings, yet their sensitivity to prompt fo

ArXiv cs.AI 📄 Paper ⚡ AI Lesson 11h ago

Do Neurons Dream of Primitive Operators? Wake-Sleep Compression Rediscovers Schank's Event Semantics

arXiv:2603.25975v1 Announce Type: cross Abstract: We show that they do. Schank's conceptual dependency theory proposed that all events decompose into primitive

ArXiv cs.AI 🧠 Large Language Models 📄 Paper ⚡ AI Lesson 11h ago

Policy-Guided World Model Planning for Language-Conditioned Visual Navigation

arXiv:2603.25981v1 Announce Type: cross Abstract: Navigating to a visually specified goal given natural language instructions remains a fundamental challenge in

ArXiv cs.AI 📄 Paper ⚡ AI Lesson 11h ago

Longitudinal Boundary Sharpness Coefficient Slopes Predict Time to Alzheimer's Disease Conversion in Mild Cognitive Impairment: A Survival Analysis Using the ADNI Cohort

arXiv:2603.26007v1 Announce Type: cross Abstract: Predicting whether someone with mild cognitive impairment (MCI) will progress to Alzheimer's disease (AD) is c

ArXiv cs.AI 🧠 Large Language Models 📄 Paper ⚡ AI Lesson 11h ago

FairLLaVA: Fairness-Aware Parameter-Efficient Fine-Tuning for Large Vision-Language Assistants

arXiv:2603.26008v1 Announce Type: cross Abstract: While powerful in image-conditioned generation, multimodal large language models (MLLMs) can display uneven pe

ArXiv cs.AI 👁️ Computer Vision 📄 Paper ⚡ AI Lesson 11h ago

VLAgeBench: Benchmarking Large Vision-Language Models for Zero-Shot Human Age Estimation

arXiv:2603.26015v1 Announce Type: cross Abstract: Human age estimation from facial images represents a challenging computer vision task with significant applica

ArXiv cs.AI 📄 Paper ⚡ AI Lesson 11h ago

Unlabeled Cross-Center Automatic Analysis for TAAD: An Integrated Framework from Segmentation to Clinical Features

arXiv:2603.26019v1 Announce Type: cross Abstract: Type A Aortic Dissection (TAAD) is a life-threatening cardiovascular emergency that demands rapid and precise

ArXiv cs.AI 🖌️ UI/UX Design 📄 Paper ⚡ AI Lesson 11h ago

Designing Fatigue-Aware VR Interfaces via Biomechanical Models

arXiv:2603.26031v1 Announce Type: cross Abstract: Prolonged mid-air interaction in virtual reality (VR) causes arm fatigue and discomfort, negatively affecting

ArXiv cs.AI 🧠 Large Language Models 📄 Paper ⚡ AI Lesson 11h ago

H-Node Attack and Defense in Large Language Models

arXiv:2603.26045v1 Announce Type: cross Abstract: We present H-Node Adversarial Noise Cancellation (H-Node ANC), a mechanistic framework that identifies, exploi

ArXiv cs.AI 📄 Paper ⚡ AI Lesson 11h ago

Seeing Like Radiologists: Context- and Gaze-Guided Vision-Language Pretraining for Chest X-rays

arXiv:2603.26049v1 Announce Type: cross Abstract: Despite recent advances in medical vision-language pretraining, existing models still struggle to capture the

ArXiv cs.AI 📄 Paper ⚡ AI Lesson 11h ago

Bridging Pixels and Words: Mask-Aware Local Semantic Fusion for Multimodal Media Verification

arXiv:2603.26052v1 Announce Type: cross Abstract: As multimodal misinformation becomes more sophisticated, its detection and grounding are crucial. However, cur

ArXiv cs.AI 🧠 Large Language Models 📄 Paper ⚡ AI Lesson 11h ago

MuDD: A Multimodal Deception Detection Dataset and GSR-Guided Progressive Distillation for Non-Contact Deception Detection

arXiv:2603.26064v1 Announce Type: cross Abstract: Non-contact automatic deception detection remains challenging because visual and auditory deception cues often

ArXiv cs.AI 👁️ Computer Vision 📄 Paper ⚡ AI Lesson 11h ago

R-PGA: Robust Physical Adversarial Camouflage Generation via Relightable 3D Gaussian Splatting

arXiv:2603.26067v1 Announce Type: cross Abstract: Physical adversarial camouflage poses a severe security threat to autonomous driving systems by mapping advers

ArXiv cs.AI 🧠 Large Language Models 📄 Paper ⚡ AI Lesson 11h ago

When Identities Collapse: A Stress-Test Benchmark for Multi-Subject Personalization

arXiv:2603.26078v1 Announce Type: cross Abstract: Subject-driven text-to-image diffusion models have achieved remarkable success in preserving single identities

ArXiv cs.AI 🧠 Large Language Models 📄 Paper ⚡ AI Lesson 11h ago

Selective Deficits in LLM Mental Self-Modeling in a Behavior-Based Test of Theory of Mind

arXiv:2603.26089v1 Announce Type: cross Abstract: The ability to represent oneself and others as agents with knowledge, intentions, and belief states that guide

ArXiv cs.AI 📄 Paper ⚡ AI Lesson 11h ago

Dynamic Tokenization via Reinforcement Patching: End-to-end Training and Zero-shot Transfer

arXiv:2603.26097v1 Announce Type: cross Abstract: Efficiently aggregating spatial or temporal horizons to acquire compact representations has become a unifying

ArXiv cs.AI 🧠 Large Language Models 📄 Paper ⚡ AI Lesson 11h ago

A Human-Inspired Decoupled Architecture for Efficient Audio Representation Learning

arXiv:2603.26098v1 Announce Type: cross Abstract: While self-supervised learning (SSL) has revolutionized audio representation, the excessive parameterization a

ArXiv cs.AI 🧠 Large Language Models 📄 Paper ⚡ AI Lesson 11h ago

"Oops! ChatGPT is Temporarily Unavailable!": A Diary Study on Knowledge Workers' Experiences of LLM Withdrawal

arXiv:2603.26099v1 Announce Type: cross Abstract: LLMs have become deeply embedded in knowledge work, raising concerns about growing dependency and the potentia

📰 AI News