TY - JOUR
AU - He, Zhimin
AU - Li, Lvzhou
AU - Zheng, Shenggen
AU - Zou, Xiangfu
AU - Situ, Haozhen
PY - 2019
DA - 2019/09/30
TI - Quantum speedup for pool-based active learning
JO - Quantum Information Processing
SP - 345
VL - 18
IS - 11
AB - Active learning aims to select the most informative samples to train an accurate classifier with minimum cost of labeling. It is widely used in many machine learning systems, where there are a large amount of unlabeled data, but it is difficult or expensive to obtain their labels due to the involvement of human efforts. However, active learning is time-consuming, particularly for the applications those have a great number of unlabeled samples, such as image retrieval, text mining and speech recognition. Thus, it is crucial to speed up the active learning algorithm. In this paper, we propose a quantum version of active learning algorithm, which converts a classical active learning to its quantum counterpart. We focus on the pool-based active learning, which is one of the most popular branches of active learning. The proposed quantum active learning algorithm can achieve quadratic speedup over the classical pool-based active learning.
SN - 1573-1332
UR - https://doi.org/10.1007/s11128-019-2460-x
DO - 10.1007/s11128-019-2460-x
ID - He2019
ER -