Infothek
This page contains automatically translated content.
New paper accepted for the Workshop Interactive Adaptive Learning (IAL), ECML PKDD 2023
The paper titled "Role of Hyperparameters in Deep Active Learning" by Denis Huseljic, Marek Herde, Paul Hahn, and Bernhard Sick addresses the often overlooked problem of selecting appropriate training hyperparameters (HPs), such as learning rate, that play a crucial role in how deep neural networks learn during each training cycle. This study shows that optimizing these hyperparameters can help reduce the performance differences between different DAL strategies.
News
New paper accepted for the Workshop Interactive Adaptive Learning (IAL), ECML PKDD 2023
The paper titled "Role of Hyperparameters in Deep Active Learning" by Denis Huseljic, Marek Herde, Paul Hahn, and Bernhard Sick addresses the often overlooked problem of selecting appropriate training hyperparameters (HPs), such as learning rate, that play a crucial role in how deep neural networks learn during each training cycle. This study shows that optimizing these hyperparameters can help reduce the performance differences between different DAL strategies.
Dates
New paper accepted for the Workshop Interactive Adaptive Learning (IAL), ECML PKDD 2023
The paper titled "Role of Hyperparameters in Deep Active Learning" by Denis Huseljic, Marek Herde, Paul Hahn, and Bernhard Sick addresses the often overlooked problem of selecting appropriate training hyperparameters (HPs), such as learning rate, that play a crucial role in how deep neural networks learn during each training cycle. This study shows that optimizing these hyperparameters can help reduce the performance differences between different DAL strategies.