Results for "data-driven"
US approval process for medical AI devices.
Software regulated as a medical device.
Unequal performance across demographic groups.
Fabrication of cases or statutes by LLMs.
Ultra-low-latency algorithmic trading.
Simulating adverse scenarios.
Predicting borrower default risk.
Quantifying financial risk.
AI discovering new compounds/materials.
Incremental capability growth.
Fast approximation of costly simulations.
Stored compute or algorithms enabling rapid jumps.
Updating a pretrained model’s weights on task-specific data to improve performance or adapt style/behavior.
A measure of a model class’s expressive capacity based on its ability to shatter datasets.
Measures a model’s ability to fit random noise; used to bound generalization error.
Inferring the agent’s internal state from noisy sensor data.