Chelsea Finn (@chelseabfinn)

연구 결과: 일반적으로 데이터 다양성이 많을수록 좋다고 예상하지만, 크로스-엠바디먼트(cross-embodiment) 전이에는 서로 짝지어진(paired) 서로 다른 엠바디먼트 데이터가 다양성 증가보다 더 유리하다는 발견을 보고합니다. 관련 웹페이지·코드(data-analogies)와 arXiv 논문 링크 제공.

https://x.com/chelseabfinn/status/2033339098694631427

#transferlearning #crossembodiment #datasets #robotics

Vision LLMs may skip training data, but you still need test images.

#ai #vision #datasets

‘Thousands of authors including Kazuo Ishiguro, Philippa Gregory and Richard Osman have published an “empty” #book to protest against #AI firms using their work without permission.’

https://www.theguardian.com/technology/2026/mar/10/thousands-authors-publish-empty-book-protest-ai-work-copyright

Yet #copyright itself has long been criticised as part of broader systems of enclosure and #SettlerViolence. So the assertion of copyright is not a victimless crime, any more than is the training of AI #chatbots and image generators on vast #datasets (often scraped without permission from the open web, digital repositories and shadow or #pirate libraries containing copyrighted books).

So what exactly is being defended here? Do the authors protesting against the training of AI really not know the long history of critique of copyright? Or do they know perfectly and are just too selfish and are profiting too much from it themselves to want to challenge it or think of something different?

#DefundCulture

Earth’s Future: Insights From Managed Retreat Projects in Europe. “Here, we present a data set of European managed retreat cases, based on a multilingual review of academic and gray literature, as well as media articles. We found 44 implemented or planned cases of managed retreat across the continent, spanning 11 countries, ranging from the relocation of individual assets to more than 1,500 […]

Datasets for Reconstructing Visual Perception from Brain Data

https://github.com/seelikat/neuro-visual-reconstruction-dataset-index

#HackerNews #Datasets #VisualPerception #BrainData #Neuroscience #Research #MachineLearning

Scientific Data: WorldMove, a global open data for human mobility . “High-quality human mobility data is crucial for applications such as urban planning, transportation management, and public health, yet its collection is often hindered by privacy concerns and data scarcity, particularly in less-developed regions. To address this challenge, we introduce WorldMove, a large-scale synthetic […]

NASA: New Volunteer Data from 143 Observatories Unveils the 2024 Total Solar Eclipse . “This data, which includes 52,469 total photographs uploaded by project volunteers, is now live: https://eclipsemegamovie.org/database. The data include contributions from 143 unique, mobile, volunteer-led ‘observatories’ – people with cameras charged with taking precise images of the eclipse, taking extra […]

153M+ research outputs in the #OpenAIREGraph are linked to #datasets & #software.

This growing web of connections makes it easier to see how knowledge is built across publications, data, & code, not just the final paper.

Explore these connections with the #GraphAPI or in #OpenAIREEXPLORE
🔗 API https://shorturl.at/oRotk
🔗 EXPLORE https://shorturl.at/RIZoh

ITmedia AI＋ (@itm_aiplus)

AI 학습 목적의 불법 복제물(해적판) 수집·이용이 저작권법 위반에 해당하는지에 대한 법적 검토를 소개. 변호사 柿沼太一(카키누마 타이치)의 견해를 다루며, AI 학습 데이터와 저작권 충돌 문제를 논의하는 내용임을 알림.

https://x.com/itm_aiplus/status/2023926708009922839

#copyright #ai #law #datasets #legal