- Launchdeepmind.google
AI-Generated Video Detection via Perceptual Straightening
DeepMind researchers have developed a novel approach called ReStraV (Representation Straightening Video) to detect AI-generated videos. This method leverages the 'perceptual straightening' hypothesis,…
- Launchdeepmind.google
EmbeddingGemma: Powerful and Lightweight Text Representations
DeepMind introduces EmbeddingGemma, a lightweight, open text embedding model based on the Gemma 3 language model family. The model uses an innovative training recipe that captures knowledge from large…
- Launchdeepmind.google
Video models are zero-shot learners and reasoners
DeepMind's Veo 3 model demonstrates remarkable zero-shot capabilities, solving a wide range of tasks it wasn't explicitly trained for, such as object segmentation, edge detection, image editing, under…
- Technicaldeepmind.google
Improving cosmological reach of LIGO usingDeep Loop Shaping
DeepMind researchers have developed a reinforcement learning (RL) based controller to improve the low-frequency band disturbance rejection in gravitational wave (GW) detectors like LIGO. This advancem…
- Launchai.google.dev
Gemma models overview
Google has introduced Gemma, a family of lightweight, state-of-the-art open models derived from the same technology used to create the Gemini models. Developed by Google DeepMind and other teams acros…
DeepMind 在 2025年9月 发布了 5 项受追踪的更新。 较 2025年7月 增长 150%。 主要主题是 product launches(4)。 亮点:“AI-Generated Video Detection via Perceptual Straightening”(2025年9月29日)。
信号分类 — 2025年9月
4
product launches
1
technical updates
浏览 DeepMind 的其他月份