← 전체 목록

2026-10-04 AI 리서치 브리핑

최신 VLM, sLLM, on-device AI 논문과 연구 블로그를 한눈에 정리합니다. 중복 기사 방지를 위해 URL 기준으로 추적합니다.

총 18건 요약 자동 생성

VLM 업데이트

멀티모달 비전-언어 모델의 최신 논문과 리더보드 변화

ROWBench: Do Video Models Render What the Program Specifies?

Paper arXiv cs.CV (recent)

Programmable world models separate executable dynamics from visual generation, offering a promising foundation for next-generation game engines. However, their visual adherence to explicit rules and interactions remains insufficiently evaluated. Existing benchmarks assess visual quality, controllability, and instruction or physical adherence, but rarely test fidelity to fine-grained, program-specified world events. We introduce PROWBench, comprising 170 programmatically constructed episodes and 600 proxy videos covering diverse scenes and interactions. PROWBench logs entity states and timestamped events, including those outside the camera's field of view, as replayable world records, from which it renders synchronized views and proxy representations. This enables generated videos to be checked against the observable consequences of program execution. An extensible framework constructs scenes, controls behaviors, and can render each camera view in different representations, such as coarse 3D, and bounding boxes. The benchmark covers first- and third-person perspectives, with synchronized multi-view observations available for a subset of episodes. Grounded in these records, PROWBench evaluates entity control, long-horizon memory, and, with two VLM-based metrics, Logic-Render Alignment and Interaction Success Rate, adherence to the prescribed timeline and the visual realization of timestamped engine-recorded events.

원문 보기

sLLM 트렌드

경량화·효율화를 위한 스몰 LLM 연구

On-Policy or Off-Policy Learning? A Systematic Study of Distillation Dynamics

Paper Hugging Face Papers

On-Policy or Off-Policy Learning? A Systematic Study of Distillation Dynamics에 관한 최근 업데이트입니다. 자세한 내용은 원문 링크에서 확인할 수 있습니다.

원문 보기

One Basis to Animate Them All: Gaussian Blendshape Distillation for Real-Time Avatars

Paper arXiv cs.CV (recent)

3D Gaussian avatars support fast rendering, however, their real-time animation is often challenged by the costly neural inference. We address this bottleneck and show that the animation of pretrained avatar models can be closely approximated by a linear combination of identity-independent blendshapes. Building on this finding, we introduce GALA (Gaussian Animation via Linear Approximation), a distillation method that replaces per-frame heavy neural decoding with a shallow coefficient predictor and a linear blend. To improve fidelity and reduce memory requirements, we propose to construct the basis using block-local PCA under a rendering-aware metric and a memory budget. Our method learns a shallow MLP network to predict blendshape coefficients and applies to various animation architectures without retraining original models. We validate GALA by accelerating the inference of three distinct avatar models for 3D animation of facial expressions and full-bodies with clothing dynamics. Across these models, our distillation generalizes to held-out identities and reduces CPU animation cost by up to three orders of magnitude while preserving most of the rendering quality. Excellent results of our method confirm the shared linear structure of learned avatar representations and enable highly efficient and accurate animation at frame rates reaching up to 60fps on mobile devices. Project page: https://ramazan793.github.io/gala/

원문 보기

On-Device AI

디바이스 내 추론 및 엣지 최적화 동향

Embedding Prediction Helps Image Generation

Paper arXiv cs.CV (recent)

In diffusion transformers, a class label or a text prompt is embedded once, and the same condition is reused at every denoising step. We ask whether predicted embeddings can serve as this condition instead. Next-Embedding Predictive Autoregression (NEPA) trains a Transformer to predict the next continuous embedding in a sequence. In generation, the clean image follows the noisy image, so its embeddings are the next embeddings after the condition and the noisy image. We train a NEPA model to predict them all at once with Multi-Embedding Prediction, and in Embedding Conditioned Generation, a DiT generator is conditioned on these predictions, recomputed at every denoising step, so the conditioning signal adapts to the current noisy state. Experiments on class-conditional ImageNet $256\times256$ study the condition of the generator, the design of Multi-Embedding Prediction, and the scaling of both models. The NEPA model adds a second network to every sampling step; with it, and combined with REPA, our final model, NEPA-DiT-XL, reaches an FID of 1.32 using about a third of the training compute of REPA.

원문 보기

October 2, 2026 Toward provably private learning from federated data Mobile Systems · Security, Privacy and Abuse Prevention · Software Systems & Engineering

News Google Research Blog

October 2, 2026 Toward provably private learning from federated data Mobile Systems · Security, Privacy and Abuse Prevention · Software Systems & Engineering에 관한 최근 업데이트입니다. 자세한 내용은 원문 링크에서 확인할 수 있습니다.

원문 보기

AI 뉴스 & 리서치

기업/연구기관의 주요 발표와 블로그 업데이트

OneStreamer: Unifying Perception, Memory, and Proactive Response in Streaming Video Interaction

Paper Hugging Face Papers

OneStreamer: Unifying Perception, Memory, and Proactive Response in Streaming Video Interaction에 관한 최근 업데이트입니다. 자세한 내용은 원문 링크에서 확인할 수 있습니다.

원문 보기

Adaptive Reward Routing: Dynamic Multi-Reward Optimization for Joint Audio-Video Diffusion via Forward-Process RL

Paper Hugging Face Papers

Adaptive Reward Routing: Dynamic Multi-Reward Optimization for Joint Audio-Video Diffusion via Forward-Process RL에 관한 최근 업데이트입니다. 자세한 내용은 원문 링크에서 확인할 수 있습니다.

원문 보기

GraphForge: Training Working Agents with Graph-Anchored Workspace Synthesis

Paper Hugging Face Papers

GraphForge: Training Working Agents with Graph-Anchored Workspace Synthesis에 관한 최근 업데이트입니다. 자세한 내용은 원문 링크에서 확인할 수 있습니다.

원문 보기

Hierarchical Continuous Diffusion Language Models

Paper Hugging Face Papers

Hierarchical Continuous Diffusion Language Models에 관한 최근 업데이트입니다. 자세한 내용은 원문 링크에서 확인할 수 있습니다.

원문 보기

Moore, Escher, Penrose: A Conformal Golden Braid

Paper arXiv cs.CV (recent)

I don't think I have ever done anything as peculiar in my life. Among other things, it shows a young man looking with interest at a print on the wall of an exhibition that features himself. How can this be? Perhaps I am not far removed from Einstein's curved universe.'' So wrote M.C. Escher about his 1956 lithograph Print Gallery. Nearly half a century later, a mathematical analysis related its geometry to an untwisted source image through a conformal power map $z \mapsto z^α$, $α\in \mathbb{C}$. Building on this construction, we use a frozen text-to-image diffusion model to generate new self-referential scenes. Prompting alone does not enforce the recursion, while a post-hoc transformation can leave structures poorly connected. Applying the transformation during sampling is also insufficient: the denoiser may "repair" the intended distortion or drift out of the prescribed geometry. We construct a generalized inverse $T^\dagger$ of the non-invertible image transformation $T$, adapted to its recursive constraint. In the idealized formulation, the Penrose identity $TT^\dagger T = T$ makes $TT^\dagger$ an idempotent projection onto geometrically admissible images. Yet denoising only the transformed image remains an out-of-distribution task, even with projection. We therefore braid denoising steps with $T$ and $T^\dagger$: source-space steps develop the untwisted scene, while transformed-space steps refine its appearance and connections in the final geometry. We generate Print Gallery-like compositions and explore further transformations. Rather than distorting a finished image, we let the scene and its distortion develop together.

원문 보기

Sphere Encoder 2

Paper arXiv cs.CV (recent)

Sphere Encoder is an autoencoder that generates images by decoding random points from a high-dimensional latent sphere. We identify two limitations of the original formulation that reduce its generation quality. First, random points concentrate near the equator relative to the pole on an encoded latent, but the training rotation never reaches this region, leaving a gap that limits one-step generation. Second, training for generation with pixel-wise reconstruction loss encourages the decoder to average over plausible images, producing blurry images that lack high-frequency details. We present Sphere Encoder 2 to address both limitations, substantially improving image generation quality while maintaining the speed and simplicity of a autoencoder. Models are released at \href{https://github.com/kaiyuyue/sphere2}{github.com/kaiyuyue/sphere2}.

원문 보기

September 29, 2026 How Diffusion Controller unifies and simplifies AI image generation Algorithms & Theory · Machine Intelligence

News Google Research Blog

September 29, 2026 How Diffusion Controller unifies and simplifies AI image generation Algorithms & Theory · Machine Intelligence에 관한 최근 업데이트입니다. 자세한 내용은 원문 링크에서 확인할 수 있습니다.

원문 보기

September 24, 2026 Automating coherent long-form video generation Generative AI · Machine Intelligence

News Google Research Blog

September 24, 2026 Automating coherent long-form video generation Generative AI · Machine Intelligence에 관한 최근 업데이트입니다. 자세한 내용은 원문 링크에서 확인할 수 있습니다.

원문 보기

September 18, 2026 MilleMiglia: A realistic instance generator for middle-mile logistics Algorithms & Theory

News Google Research Blog

September 18, 2026 MilleMiglia: A realistic instance generator for middle-mile logistics Algorithms & Theory에 관한 최근 업데이트입니다. 자세한 내용은 원문 링크에서 확인할 수 있습니다.

원문 보기

September 17, 2026 The future of practice: Enabling teachers to create learning interactives with generative UI Education Innovation · Generative AI · Machine Intelligence

News Google Research Blog

September 17, 2026 The future of practice: Enabling teachers to create learning interactives with generative UI Education Innovation · Generative AI · Machine Intelligence에 관한 최근 업데이트입니다. 자세한 내용은 원문 링크에서 확인할 수 있습니다.

원문 보기

Forecasting space weather risks on power grids

News Microsoft Research Blog

Forecasting space weather risks on power grids에 관한 최근 업데이트입니다. 자세한 내용은 원문 링크에서 확인할 수 있습니다.

원문 보기

Introducing Quine: An AI research system designed for the complexity of biology

News Microsoft Research Blog

Quine은 생물학의 복잡성을 위해 설계된 AI 연구 시스템입니다. 이 시스템에 대한 최신 업데이트는 원문 링크에서 자세히 확인할 수 있습니다.

원문 보기

One year in: How Microsoft Research Asia – Singapore is advancing research, partnership and talent for real-world impact

News Microsoft Research Blog

Microsoft Research Asia – Singapore가 1년 동안 실제 세계에 영향을 미치기 위해 연구, 파트너십 및 인재를 어떻게 발전시키고 있는지에 대한 내용입니다. 자세한 정보는 원문 링크에서 확인하실 수 있습니다.

원문 보기

참고한 소스