Add to ORKGAbstract We address problems of model misspeci\u85cation in active learning. We suppose that an investigator will sample training input points from a den-sity possibly di¤erent from that in the environment, with the intention of estimating a, possibly incorrectly speci\u85ed, linear response function and then predicting outputs at all possible values of the inputs, whether sampled or not. We derive training input densities which are asymptotically minimax robust against the losses incurred by random measurement (of the outputs) error, sampling variation (in the inputs) and biases resulting from the model mis-speci cation. Finite sample examples and simulations demonstrate the strong gains to be achieved in this manner, relative to passive l...
This paper presents a rigorous statistical analysis characterizing regimes in which active learning ...
A common belief in unbiased active learning is that, in order to capture the most informative instan...
Traditional models of active learning assume a learner can directly manipulate or query a covariate ...
This paper analyzes the potential advantages and theoretical challenges of active learning algorit...
We introduce a method of Robust Learning (‘robl’) for binary data, and propose its use in situations...
Bayesian adaptive experimental design is a form of active learning, which chooses samples to maximiz...
In many settings in practice it is expensive to obtain labeled data while unlabeled data is abundant...
This thesis presents a general discussion of active learning and adaptive sampling. In many practica...
Permission is hereby granted to the University of Alberta Libraries to reproduce single copies of th...
We propose a new active learning algorithm for parametric linear regression with random design. We p...
I study active learning in general pool-based active learning models as well noisy active learning a...
Active learning refers to the settings in which a machine learning algorithm (learner) is able to s...
Active learning (AL) is a branch of machine learning that deals with problems where unlabeled data i...
We study the robustness of active learning (AL) algorithms against prior misspecification: whether a...
We study the robustness of active learning (AL) algorithms against prior misspecification: whether a...
This paper presents a rigorous statistical analysis characterizing regimes in which active learning ...
A common belief in unbiased active learning is that, in order to capture the most informative instan...
Traditional models of active learning assume a learner can directly manipulate or query a covariate ...
This paper analyzes the potential advantages and theoretical challenges of active learning algorit...
We introduce a method of Robust Learning (‘robl’) for binary data, and propose its use in situations...
Bayesian adaptive experimental design is a form of active learning, which chooses samples to maximiz...
In many settings in practice it is expensive to obtain labeled data while unlabeled data is abundant...
This thesis presents a general discussion of active learning and adaptive sampling. In many practica...
Permission is hereby granted to the University of Alberta Libraries to reproduce single copies of th...
We propose a new active learning algorithm for parametric linear regression with random design. We p...
I study active learning in general pool-based active learning models as well noisy active learning a...
Active learning refers to the settings in which a machine learning algorithm (learner) is able to s...
Active learning (AL) is a branch of machine learning that deals with problems where unlabeled data i...
We study the robustness of active learning (AL) algorithms against prior misspecification: whether a...
We study the robustness of active learning (AL) algorithms against prior misspecification: whether a...
This paper presents a rigorous statistical analysis characterizing regimes in which active learning ...
A common belief in unbiased active learning is that, in order to capture the most informative instan...
Traditional models of active learning assume a learner can directly manipulate or query a covariate ...