AI Research news and analysis on Web Pulse: 21090 curated articles tracking the latest AI Research developments, tools, and research, updated continuously from vetted sources.
Researchers constructed UC-Bench, a human-annotated benchmark for detecting user-side implicit conflicts in Human-LLM dialogue, and found existing LLMs struggle with the task, especially when conflicts stem from implicit…
A new benchmark called Neo-Classic, introduced in arXiv paper 2609.19154v1, shows that state-of-the-art large language models lose 20 to 50 percent accuracy when moving from historical Classical Chinese Poetry to strictl…
VisKG-LM, a new method detailed in arXiv paper 2609.19158v1, compiles retrieved knowledge graphs once offline into cached images that a language model reads as read-only visual memory, decoupling graph encoding from lang…
A self-supervised method called Reflective Recovery, which turns failed reasoning trajectories into recovery training data, raised accuracy on DeepSeek-R1-Distill-Qwen-7B from 30.0% to 37.5% on AIME 2025 and from 37.6% t…
The YNU-HPCC team placed in subtask A of SemEval-2025 Task 11 with a RoBERTa-based system that scored an official ranking score of 0.44 across all languages, according to the team's arXiv paper. The team modified the mod…
Researchers introduced ReaLMem, described as the first benchmark built from authentic multi-year personal visual archives with first-person subjective annotations, to evaluate AI models on long-term multimodal personal m…
A frontend-backend architecture that lets full-duplex speech-to-speech models delegate tool calls to a text-based backend LLM achieved 92-97% tool-call recall and 81.2% accuracy in rejecting irrelevant calls in a single-…
A new arXiv paper (2609.19366v1) reports that category-specific harm representations in large language models carry safety-relevant information beyond a shared general harmfulness direction. Using activation steering wit…
TrioRAG, a graph-free multimodal retrieval-augmented generation framework, matches or outperforms graph-based systems across three benchmarks while cutting total cost and speeding per-query inference by 1.6 to 2.3 times,…
Researchers released SonoCorpus, an open ultrasound dataset of 456,963 images and 1,626,085 expert masks drawn from 53 public datasets across 24 clinical applications and 17 countries, alongside SonoBase, an interactive …
A new arXiv paper (2609.19384v1) proposes Riemannian–Lorentz Parameter Fusion (RLPF), a geometry-aware method for merging a Vision Transformer and a state-space model into a two-branch hybrid that reaches 82.37% on CIFAR…
A regression framework built on open-source Vision-Language Models reached a Spearman correlation of 0.67 with a four-model configuration when scoring Olympic diving videos from the AQA-7 benchmark, according to an arXiv…
Efficient Unified Multimodal Understanding (EUMU) won the MUMU Track of the 8th LSVOS Challenge with a final score of 17.3409, using a single shared pretrained multimodal model to jointly perform multi-concept image tagg…
ParticleSplat, a self-supervised object-centric learning method detailed in arXiv paper 2609.19463v1, decomposes scenes into latent "particles" representing semantic entities via feedforward 3D Gaussian Splatting. Buildi…
Researchers introduced SOLAR, a multimodal generative model that analyzes tomato leaf disease across six question-answering tasks by aligning visual features with task-aware language representations through a mixture-of-…
A deep-learning perception framework using the YOLOv12-m-seg model achieved a segmentation AP@0.5 of 83.7% with an inference time of 20.4 ms per image, the most favorable balance of accuracy and speed among direct segmen…
A Retrieval-Augmented Generation framework for scientific image quality assessment took 1st place in the SIQA-U understanding track of the SIQA challenge at the ICME 2026 Grand Challenges, according to arXiv paper 2609.1…
LABELSENSE-Pilot, a reproducible prototype for point-feature label placement on interactive maps, displayed 85.62 percent of labels with 2.09 percent flicker and zero collisions across five seeds on 2,500 airport coordin…
A new arXiv paper (2609.19425v1) reports that tool-augmented LLM agents hallucinate calls to tools that do not exist, and that no existing tool-selection or gating defense can reject them because a fabricated call is by …
Researchers posted arXiv:2609.19996v1, a deep architecture that jointly optimizes cost functions and a route-ranking model for customizable route planning. The system runs a multi-objective Dijkstra algorithm offline to …