Synthetic data

Foundation models and synthetic data are increasingly used in scientific applications, yet the reliability of their outputs is often uncertain. This workshop investigates how model predictions and generated data can be incorporated into statistical inference without losing trustworthy uncertainty estimates. Researchers from several disciplines will study the implications of black-box models for inference and decision-making and develop methods for assessing and improving reliability.

…The doctorate combines RGB-D data, scene understanding, synthetic data, simulation-to-reality transfer and robot manipulation, with experimental validation on physical platforms. The project is hosted by the Advanced Manufacturing Lab and includes collaboration with researchers and industrial partners. Apply through the ETH online recruitment portal. The vacancy is currently advertised with an active application route; no closing date is stated.

…and public-domain sources alongside synthetic data. The conversation examines the SYNTH dataset, gaps in common web crawls, preservation of source material and ethical choices beyond copyright status. It also covers compact deployed models, benchmark incentives, reasoning-trace access and competing approaches to national AI development. Hosted by Ravid Shwartz Ziv and Allen Roush. Watch the full conversation on the publisher’s YouTube channel.

Maissam Barkeshli uses synthetic data of adjustable complexity to study how transformers learn. Sequences from graph, hypergraph and hierarchical random walks make it possible to investigate scaling laws, sample complexity and unexpected temperature effects in controlled settings. The presentation also considers how parameterisation changes hyperparameter scaling, including the importance of scaling the embedding-layer learning rate.