TY - GEN
T1 - Topology-Preserving Class-Incremental Learning
AU - Tao, Xiaoyu
AU - Chang, Xinyuan
AU - Hong, Xiaopeng
AU - Wei, Xing
AU - Gong, Yihong
N1 - Publisher Copyright:
© 2020, Springer Nature Switzerland AG.
PY - 2020
Y1 - 2020
N2 - A well-known issue for class-incremental learning is the catastrophic forgetting phenomenon, where the network’s recognition performance on old classes degrades severely when incrementally learning new classes. To alleviate forgetting, we put forward to preserve the old class knowledge by maintaining the topology of the network’s feature space. On this basis, we propose a novel topology-preserving class-incremental learning (TPCIL) framework. TPCIL uses an elastic Hebbian graph (EHG) to model the feature space topology, which is constructed with the competitive Hebbian learning rule. To maintain the topology, we develop the topology-preserving loss (TPL) that penalizes the changes of EHG’s neighboring relationships during incremental learning phases. Comprehensive experiments on CIFAR100, ImageNet, and subImageNet datasets demonstrate the power of the TPCIL for continuously learning new classes with less forgetting. The code will be released.
AB - A well-known issue for class-incremental learning is the catastrophic forgetting phenomenon, where the network’s recognition performance on old classes degrades severely when incrementally learning new classes. To alleviate forgetting, we put forward to preserve the old class knowledge by maintaining the topology of the network’s feature space. On this basis, we propose a novel topology-preserving class-incremental learning (TPCIL) framework. TPCIL uses an elastic Hebbian graph (EHG) to model the feature space topology, which is constructed with the competitive Hebbian learning rule. To maintain the topology, we develop the topology-preserving loss (TPL) that penalizes the changes of EHG’s neighboring relationships during incremental learning phases. Comprehensive experiments on CIFAR100, ImageNet, and subImageNet datasets demonstrate the power of the TPCIL for continuously learning new classes with less forgetting. The code will be released.
KW - Class-Incremental Learning (CIL)
KW - Elastic Hebbian Graph (EHG)
KW - Topology-Preserving Class-Incremental Learning (TPCIL)
KW - Topology-Preserving Loss (TPL)
UR - https://www.scopus.com/pages/publications/85097257382
U2 - 10.1007/978-3-030-58529-7_16
DO - 10.1007/978-3-030-58529-7_16
M3 - 会议稿件
AN - SCOPUS:85097257382
SN - 9783030585280
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 254
EP - 270
BT - Computer Vision – ECCV 2020 - 16th European Conference, 2020, Proceedings
A2 - Vedaldi, Andrea
A2 - Bischof, Horst
A2 - Brox, Thomas
A2 - Frahm, Jan-Michael
PB - Springer Science and Business Media Deutschland GmbH
T2 - 16th European Conference on Computer Vision, ECCV 2020
Y2 - 23 August 2020 through 28 August 2020
ER -