A new method for surgically removing training examples from a model reveals that as datasets grow, the link between what a model learns and what it produces dissolves.
Machine-generated by The Flow from the publisher's headline and feed description — not written or checked by a human. The full article lives at MIT News AI.
arXiv:2412.00176v4 Announce Type: replace Abstract: We explore whether pre-training on datasets with paintings is necessary for a model to learn an artistic style with only a few examples. To investi...
arXiv:2607. 09705v1 Announce Type: cross Abstract: Since 2023, computer scientists have warned against model collapse -- the contamination of training sets with AI-generated outputs that progressively degrade model performance.
We’ve designed a method that encourages AIs to teach each other with examples that also make sense to humans.
arXiv:2608.29644v1 Announce Type: cross Abstract: Attributing an artwork to an artist has traditionally relied on detailed visual observations and descriptions, known as stylistic analysis in art his...
The paper introduces the CO-AID dataset, which captures systematic defects in state‑of‑the‑art text‑to‑image models when prompts involve complex composition such as multiple entities and attributes. Researchers manually curated 651 reference images across people, hand, object, and scene categories, edited ChatGPT‑generated prompts to emphasize compositional factors, and generated AI images with three T2I models. A subjective study with 29 participants produced multi‑label defect annotations, enabling training of a deep model that predicts defects and improves image generation.
arXiv:2606. 25996v1 Announce Type: cross Abstract: We introduce Autodata, a general method that enables AI agents to act as data scientists who build high quality training and evaluation data.