Adaptive Aggregation Networks for Class-Incremental Learning

By Yaoyao Liu, Bernt Schiele, Qianru Sun. 2020

284 citations recorded by the source corpus.

Class-Incremental Learning (CIL) aims to learn a classification model with the number of classes increasing phase-by-phase. An inherent problem in CIL is the stability-plasticity dilemma between the learning of old and new classes, i.e., high-plasticity models easily forget old classes, but high-stability models are weak to learn new classes. We alleviate this issue by proposing a novel network architecture called Adaptive Aggregation Networks (AANets) in which we explicitly build two types of residual blocks at each residual level (taking ResNet as the baseline architecture): a stable block and a plastic block. We aggregate the output feature maps from these two blocks and then feed the results to the next-level blocks. We adapt the aggregation weights in order to balance these two types of blocks, i.e., to balance stability and plasticity, dynamically. We conduct extensive experiments on three CIL benchmarks: CIFAR-100, ImageNet-Subset, and ImageNet, and show that many existing CIL methods can be straightforwardly incorporated into the architecture of AANets to boost their performances1.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition, Radiology, Nuclear Medicine and Imaging, Meta & Continual Learning

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Computer Vision and Pattern Recognition

In Radiology, Nuclear Medicine and Imaging

In Meta & Continual Learning

Citations in the corpus

A Comprehensive Survey of Continual Learning: Theory, Method and Application (2023, methodology): “Feature Space LUCIR [164], PODNet [100], TPCIL [403], PCL [170], AANets [270], DER [468], DDE [171], GeoDL [387], PASS [510], Co2L [57], AFC [199], SP-CIL [456], ELI [192], CwD [380], CSCCT [22], FOSTER [424], FASP [300], CLS-ER [18]”
Learnability and Algorithm for Continual Learning (2023, methodology): “…buffer data (Rusu et al., 2016; LopezPaz & Ranzato, 2017; Chaudhry et al., 2019a; Hou et al., 2019; Wu et al., 2019; Rolnick et al., 2019; Buzzega et al., 2020; Rajasegaran et al., 2020a; Liu et al., 2021; Cha et al., 2021; Yan et al., 2021; Wang et al., 2022b; Guo et al., 2022; Kim et al., 2022a).”
RMM: Reinforced Memory Management for Class-Incremental Learning (2023, methodology): “We follow the benchmark protocol used in [13, 18, 30, 31].”
FeTrIL: Feature Translation for Exemplar-Free Class-Incremental Learning (2022, methodology): “Comparison of FeTrIL with the recent AANets method [21], applied on top of EBCIL baselines which store 20 exemplars of past classes to mitigate catastrophic forgetting.”
A Theoretical Study on Solving Continual Learning (2022, background): “[36] Yaoyao Liu, Bernt Schiele, and Qianru Sun.”
Long-Tailed Class Incremental Learning (2022, methodology): “Further results on AANets [23] and DER [43] We report results for AANets (based on LUCIR) on Shuffle LT-CIL scenario (CIFAR-100 with 10- task setting).”
Novel Class Discovery without Forgetting (2022, background): “Dynamically expanding and parameter isolation methods [44,40,39,1,33] form a third class of methods to address forgetting.”
Class-Incremental Learning by Knowledge Distillation with Adaptive Feature Consolidation (2022, methodology): “Architectural approaches employ network expansion schemes [27, 41, 49] instead of using static models.”
Energy-based Latent Aligner for Incremental Learning (2022, methodology): “ELI is able to consistently improve iCaRL [41], LUCIR [20] and AANet [31] on 5 task, 10 task and 25 task setting.”
Federated Class-Incremental Learning (2022, background): “[29,50] design a self-adaptive network to balance biased predictions.”
Self-Sustaining Representation Expansion for Non-Exemplar Class-Incremental Learning (2022, neutral): “However, most of the existing methods [21, 30] assume that a certain number (e.”
Representation Compensation Networks for Continual Semantic Segmentation (2022, methodology): “Some other methods [45,57,75,76,82,91] increase the capacity of the model to have a better tradeoff of stability and plasticity, but with the cost of growing memory of the network.”
Memory Replay with Data Compression for Continual Learning (2022, positive): “…of representative old training samples (referred to as memory replay ) can generally achieve the best performance in class-incremental learning (Liu et al., 2021a; Hu et al., 2021) and in numerous other continual learning scenarios, such as audio tasks (Ehret et al., 2020), few-shot (Tao et…”
Mimicking the Oracle: An Initial Phase Decorrelation Approach for Class Incremental Learning (2021, methodology): “When performing CIL, we reserve a constant number of exemplars for each previously learned class as had been done in [8,12,19,20].”
Overcoming Catastrophic Forgetting in Incremental Few-Shot Learning by Finding Flat Minima (2021, methodology): “Incremental learning focuses on the challenging problem of continually learning to recognize new classes in new coming data without forgetting old classes [6, 7, 10, 51].”
Co-Transport for Class-Incremental Learning (2021, background): “Class-incremental learning (CIL) is now a popular topic in the machine learning community [6, 9, 15, 25, 39, 47, 62, 68, 70].”
Few-Shot Incremental Learning with Continually Evolved Classifiers (2021, background): “Incremental learning (IL) is an active machine learning task that aims to learn new knowledge continually without forgetting [3, 4, 7, 25].”
Distilling Causal Effect of Data in Class-Incremental Learning (2021, background): “Incremental learning aims to continuously learn by accumulating past knowledge [2, 19, 24].”