TY - GEN
T1 - Graph-based active learning
T2 - 2016 IEEE Global Conference on Signal and Information Processing, GlobalSIP 2016
AU - Jun, Kwang Sung
AU - Nowak, Robert
N1 - Funding Information:
This work was partially supported by the National Science Foundation grants CCF-1218189 and IIS-1447449 and by MURI grant ARMY W911NF-15-1-0479.
Publisher Copyright:
© 2016 IEEE.
PY - 2017/4/19
Y1 - 2017/4/19
N2 - In graph-based active learning, algorithms based on expected error minimization (EEM) have been popular and yield good empirical performance. The exact computation of EEM optimally balances exploration and exploitation. In practice, however, EEM-based algorithms employ various approximations due to the computational hardness of exact EEM. This can result in a lack of either exploration or exploitation, which can negatively impact the effectiveness of active learning. We propose a new algorithm TSA (Two-Step Approximation) that balances between exploration and exploitation efficiently while enjoying the same computational complexity as existing approximations. Finally, we empirically show the value of balancing between exploration and exploitation in both toy and real-world datasets where our method outperforms several state-of-the-art methods.
AB - In graph-based active learning, algorithms based on expected error minimization (EEM) have been popular and yield good empirical performance. The exact computation of EEM optimally balances exploration and exploitation. In practice, however, EEM-based algorithms employ various approximations due to the computational hardness of exact EEM. This can result in a lack of either exploration or exploitation, which can negatively impact the effectiveness of active learning. We propose a new algorithm TSA (Two-Step Approximation) that balances between exploration and exploitation efficiently while enjoying the same computational complexity as existing approximations. Finally, we empirically show the value of balancing between exploration and exploitation in both toy and real-world datasets where our method outperforms several state-of-the-art methods.
KW - Active learning
KW - Graph-based learning
KW - Machine learning
KW - Probabilistic model
KW - Semi-supervised learning
UR - https://www.scopus.com/pages/publications/85019252356
UR - https://www.scopus.com/pages/publications/85019252356#tab=citedBy
U2 - 10.1109/GlobalSIP.2016.7906056
DO - 10.1109/GlobalSIP.2016.7906056
M3 - Conference contribution
AN - SCOPUS:85019252356
T3 - 2016 IEEE Global Conference on Signal and Information Processing, GlobalSIP 2016 - Proceedings
SP - 1325
EP - 1329
BT - 2016 IEEE Global Conference on Signal and Information Processing, GlobalSIP 2016 - Proceedings
PB - Institute of Electrical and Electronics Engineers Inc.
Y2 - 7 December 2016 through 9 December 2016
ER -