Deep active learning: Unified and principled method for query and training
In this paper, we are proposing a unified and principled method for both the querying and
training processes in deep batch active learning. We are providing theoretical insights from
the intuition of modeling the interactive procedure in active learning as distribution matching,
by adopting the Wasserstein distance. As a consequence, we derived a new training loss
from the theoretical analysis, which is decomposed into optimizing deep neural network
parameters and batch query selection through alternative optimization. In addition, the loss …
training processes in deep batch active learning. We are providing theoretical insights from
the intuition of modeling the interactive procedure in active learning as distribution matching,
by adopting the Wasserstein distance. As a consequence, we derived a new training loss
from the theoretical analysis, which is decomposed into optimizing deep neural network
parameters and batch query selection through alternative optimization. In addition, the loss …
[PDF][PDF] Deep Active Learning: Unified and Principled Method for Query and Training [Supplementary Material]
Theorem 1. Supposing D is the data generation distribution and Q is the querying
distribution, if the loss l is symmetric, L-Lipschitz;∀ h∈ H is at most H-Lipschitz function and
underlying labeling function h⋆ is φ (λ)-(D, Q) Joint Probabilistic Lipschitz, then the expected
risk wrt D can be upper bounded by:
distribution, if the loss l is symmetric, L-Lipschitz;∀ h∈ H is at most H-Lipschitz function and
underlying labeling function h⋆ is φ (λ)-(D, Q) Joint Probabilistic Lipschitz, then the expected
risk wrt D can be upper bounded by:
Showing the best results for this search. See all results