AllLLMRoboticsStartupsResearchEthicsBusinessHardwareAutonomousMedicineCreative

All news

10709
A USTC Team's Startup Raised Tens of Millions Three Months After Founding
Robotics·Jiqizhixin (机器之心)

A USTC Team's Startup Raised Tens of Millions Three Months After Founding

Oct 9, 2026·2 min
正行创新 Unveils Physical AI for Retail: Robots Working the Shelves 24/7
Robotics

正行创新 Unveils Physical AI for Retail: Robots Working the Shelves 24/7

Oct 9, 2026·2 min
Doubao adds an infinite canvas for creative work and a 2.1 Lite model
Business

Doubao adds an infinite canvas for creative work and a 2.1 Lite model

Oct 9, 2026·2 min
Video Generation With an Expression Gallery Beats Baseline Models on Identity Similarity
Research

Video Generation With an Expression Gallery Beats Baseline Models on Identity Similarity

Oct 9, 2026·2 min
Qwen3-1.7B Fine-Tuned on Raw, Uncaptioned Video Gains 5.1 Points on Image Benchmarks
Research·arXiv cs.CV

Qwen3-1.7B Fine-Tuned on Raw, Uncaptioned Video Gains 5.1 Points on Image Benchmarks

Researchers trained Qwen3-1.7B on YT-Temporal-1B video with no text or captions: +5.1 points on ten image benchmarks, +2.9 on video, and no drop in text skills.

Oct 9, 2026·2 min
PCAsplat Cleans Up Gaussian Splat "Floaters" With Local PCA on Three Benchmarks
Research·arXiv cs.CV

PCAsplat Cleans Up Gaussian Splat "Floaters" With Local PCA on Three Benchmarks

The PCAsplat regularizer uses differentiable local PCA to pull Gaussians toward the surface and remove floater artifacts on DTU, Tanks and Temples, and NeRF…

Oct 9, 2026·2 min
Researchers Extend DINOv2's CLS Token with Regions of Interest for Image Retrieval
Research·arXiv cs.CV

Researchers Extend DINOv2's CLS Token with Regions of Interest for Image Retrieval

An arXiv preprint adds local ROI tokens to the [CLS] and four register tokens of DINOv2-reg and searches over them ColBERT-style. No boxes, detectors, or stored…

Oct 9, 2026·2 min
Omni-Diffusion-Distill speeds up multimodal dLLMs by up to 21x: 8 steps instead of 128
Research·arXiv cs.CV

Omni-Diffusion-Distill speeds up multimodal dLLMs by up to 21x: 8 steps instead of 128

Researchers compressed a unified diffusion LLM: images generate in 8 steps instead of 128, and image understanding takes 64 steps instead of 512. Speedups are 18.2x…

Oct 9, 2026·2 min
Top themes
01Fluid-Gen-Zero hands water physics to a simulator and cuts fluid flow error by 84%Oct 9, 2026·2 min02LVSPM Beats VGGT at Camera Pose Estimation on 16–256 Frames and Doesn't Break Down on Large ScenesOct 9, 2026·3 min03TopoGPU Computes Persistent Homology 53x Faster on AverageOct 9, 2026·2 min04GHARP animates a 3D head in real time: 13x faster on A100Oct 9, 2026·2 min05SAVER Teaches Video Models to Reason From a Few Frames, Using 1,250 ExamplesOct 9, 2026·2 min
Distilled Diffusion Models Learn to Forget Unwanted Concepts via DPO
Research·arXiv cs.CV

Distilled Diffusion Models Learn to Forget Unwanted Concepts via DPO

Oct 9, 2026·2 min
SPLIT-RL Teaches VLMs to See and Reason Separately: +1.4–6.1 pp Over GRPO
Research·arXiv cs.CV

SPLIT-RL Teaches VLMs to See and Reason Separately: +1.4–6.1 pp Over GRPO

Oct 9, 2026·2 min
Flow matching: velocity scaling cuts FID from 28.0 to 12.2 on ImageNet-256
Research·arXiv cs.CV

Flow matching: velocity scaling cuts FID from 28.0 to 12.2 on ImageNet-256

Oct 9, 2026·2 min
VICO Trains VLMs on Rewritable Images: Up to +5.0% on New Tasks
Research·arXiv cs.CV

VICO Trains VLMs on Rewritable Images: Up to +5.0% on New Tasks

Oct 9, 2026·2 min
DOGS Beats 9 Baseline Methods at Logo Generation Across Two Renderers
Research·arXiv cs.CV

DOGS Beats 9 Baseline Methods at Logo Generation Across Two Renderers

DOGS drops prompt rewriting: a GFlowNet samples a logo design graph and penalizes similarity to existing marks. The authors compared it with nine baseline methods.

Oct 9, 2026·2 min
MESSENGER cuts 3D motion forecasting error by 71.6% on nuScenes
Robotics·arXiv cs.CV

MESSENGER cuts 3D motion forecasting error by 71.6% on nuScenes

Researchers propose MESSENGER: scene flow estimation with a memory buffer and autoregression. On nuScenes, long-term EPE3D error fell 71.6%; on Argoverse 2, 67.7%.

Oct 9, 2026·2 min
LinSlot Teaches a Neural Network to Break a Scene Into Objects and Their Attributes Without Labels
Research·arXiv cs.CV

LinSlot Teaches a Neural Network to Break a Scene Into Objects and Their Attributes Without Labels

The new LinSlot architecture finds objects and their attributes in images without labels, building on the linear representation hypothesis. The authors claim DCI gains on several datasets.

Oct 9, 2026·2 min
OcuFlow Removes Eyeglass Glare in One Step and Wins 67.32% of Votes in a Blind Test
Research·arXiv cs.CV

OcuFlow Removes Eyeglass Glare in One Step and Wins 67.32% of Votes in a Blind Test

Oct 9, 2026·3 min
Researchers release VR180 dataset shot on Blackmagic cameras: 1211 stereo samples
Research·arXiv cs.CV

Researchers release VR180 dataset shot on Blackmagic cameras: 1211 stereo samples

Oct 9, 2026·2 min
SLVR Adds 14.2 Points on BLINK Relation to Qwen2.5-VL-7B Without Text CoT
Research·arXiv cs.CV

SLVR Adds 14.2 Points on BLINK Relation to Qwen2.5-VL-7B Without Text CoT

Oct 9, 2026·2 min
Google Releases a Workplace AI Agent for Enterprises, but Pricing and Integrations Remain Open Questions
Business·AI Business

Google Releases a Workplace AI Agent for Enterprises, but Pricing and Integrations Remain Open Questions

Oct 9, 2026·2 min
Researchers Prove Why Adversarial Training Makes Saliency Maps Sparse
Research·arXiv cs.LG

Researchers Prove Why Adversarial Training Makes Saliency Maps Sparse

Oct 9, 2026·2 min
Distilling ECG into PPG Memorized Identity: Fixed-Effects More Than Doubled the State Signal
Medicine

Distilling ECG into PPG Memorized Identity: Fixed-Effects More Than Doubled the State Signal

Oct 9, 2026·3 min
Subliminal learning passed a backdoor through numbers: 23.5% of replies in French
Research

Subliminal learning passed a backdoor through numbers: 23.5% of replies in French

Oct 9, 2026·2 min
PXtal Beats Baseline Models on Six Benchmarks, Four of Them Zero-Shot
Research

PXtal Beats Baseline Models on Six Benchmarks, Four of Them Zero-Shot

Oct 9, 2026·2 min
Nullify Erases Knowledge From LLMs Without Retraining, Using Null-Space Vectors
Research·arXiv cs.LG

Nullify Erases Knowledge From LLMs Without Retraining, Using Null-Space Vectors

Nullify makes an LLM "forget" private data at inference time: the weights stay unchanged, and forgetting quality on TOFU and MUSE matches baseline methods.

Oct 9, 2026·2 min
Multimodal Alzheimer's AI model reaches 0.935 AUROC on 2649 patients
Medicine·arXiv cs.LG

Multimodal Alzheimer's AI model reaches 0.935 AUROC on 2649 patients

A new framework combines MRI, clinical transformers and ODE-GRU. On ADNI, OASIS-2 and MIRIAD, diagnostic AUROC is 0.935, but on the external OASIS-3 cohort it drops…

Oct 9, 2026·2 min