Acquisition Function
A scalar function in Bayesian optimization used to determine the next evaluation point, transforming the posterior distribution into a decision. Common types include GP-UCB, Expected Improvement (EI), and Thompson sampling.
Timeline (last 90 days)
When high-fidelity data are scarce, the posterior uncertainty of the residual function is large, making the acquisition function's decisions unreliable.
The performance of Bayesian optimization depends heavily on the configuration combination of surrogate model and acquisition function, and no universally optimal configuration exists.
Acquisition functions in Bayesian optimization (such as EI, UCB, and Thompson sampling) directly rely on the predicted mean and uncertainty output by the surrogate model; if the uncertainty calibration of TFMs is poor, the acquisition function may bias toward over-exploration or premature convergence.
All Facts (3)
When high-fidelity data are scarce, the posterior uncertainty of the residual function is large, making the acquisition function's decisions unreliable.
50%UnverifiedThe performance of Bayesian optimization depends heavily on the configuration combination of surrogate model and acquisition function, and no universally optimal configuration exists.
50%UnverifiedAcquisition functions in Bayesian optimization (such as EI, UCB, and Thompson sampling) directly rely on the predicted mean and uncertainty output by the surrogate model; if the uncertainty calibration of TFMs is poor, the acquisition function may bias toward over-exploration or premature convergence.
50%