# Preserving Locality in Vision Transformers for Class Incremental Learning

arXiv: https://arxiv.org/abs/2304.06971 · HTML: https://mlsanity.com/paper/2304.06971/

## Key facts

- Cited by 2 papers in the map (data through 14 Oct 2026).
- Builds on 27 of its 35 references that are in the map.
- First posted on arXiv 14 Apr 2023.

## Builds on (most cited first, 27 of 27)

- [Deep Residual Learning for Image Recognition](https://mlsanity.com/paper/1512.03385/) (2015)
- [mixup: Beyond Empirical Risk Minimization](https://mlsanity.com/paper/1710.09412/) (2017)
- [Training data-efficient image transformers & distillation through attention](https://mlsanity.com/paper/2012.12877/) (2020)
- [Learning without Forgetting](https://mlsanity.com/paper/1606.09282/) (2016)
- [iCaRL: Incremental Classifier and Representation Learning](https://mlsanity.com/paper/1611.07725/) (2016)
- [Visual Prompt Tuning](https://mlsanity.com/paper/2203.12119/) (2022)
- [How transferable are features in deep neural networks?](https://mlsanity.com/paper/1411.1792/) (2014)
- [Continual Learning with Deep Generative Replay](https://mlsanity.com/paper/1705.08690/) (2017)
- [Learning to Prompt for Continual Learning](https://mlsanity.com/paper/2112.08654/) (2021)
- [Quantifying Attention Flow in Transformers](https://mlsanity.com/paper/2005.00928/) (2020)
- [Riemannian Walk for Incremental Learning: Understanding Forgetting and Intransigence](https://mlsanity.com/paper/1801.10112/) (2018)
- [Large Scale Incremental Learning](https://mlsanity.com/paper/1905.13260/) (2019)
- [DualPrompt: Complementary Prompting for Rehearsal-free Continual Learning](https://mlsanity.com/paper/2204.04799/) (2022)
- [ConViT: Improving Vision Transformers with Soft Convolutional Inductive Biases](https://mlsanity.com/paper/2103.10697/) (2021)
- [DER: Dynamically Expandable Representation for Class Incremental Learning](https://mlsanity.com/paper/2103.16788/) (2021)
- [On the Relationship between Self-Attention and Convolutional Layers](https://mlsanity.com/paper/1911.03584/) (2019)
- [DyTox: Transformers for Continual Learning with DYnamic TOken eXpansion](https://mlsanity.com/paper/2111.11326/) (2021)
- [S-Prompts Learning with Pre-trained Transformers: An Occam's Razor for Domain Incremental Learning](https://mlsanity.com/paper/2207.12819/) (2022)
- [FOSTER: Feature Boosting and Compression for Class-Incremental Learning](https://mlsanity.com/paper/2204.04662/) (2022)
- [Class-Incremental Learning: A Survey](https://mlsanity.com/paper/2302.03648/) (2023)
- [Revisiting Class-Incremental Learning with Pre-Trained Models: Generalizability and Adaptivity are All You Need](https://mlsanity.com/paper/2303.07338/) (2023)
- [A Model or 603 Exemplars: Towards Memory-Efficient Class-Incremental Learning](https://mlsanity.com/paper/2205.13218/) (2022)
- [GCR: Gradient Coreset Based Replay Buffer Selection For Continual Learning](https://mlsanity.com/paper/2111.11210/) (2021)
- [Few-Shot Class-Incremental Learning by Sampling Multi-Phase Tasks](https://mlsanity.com/paper/2203.17030/) (2022)
- [RMM: Reinforced Memory Management for Class-Incremental Learning](https://mlsanity.com/paper/2301.05792/) (2023)
- [Essentials for Class Incremental Learning](https://mlsanity.com/paper/2102.09517/) (2021)
- [Mimicking the Oracle: An Initial Phase Decorrelation Approach for Class Incremental Learning](https://mlsanity.com/paper/2112.04731/) (2021)

## Built on by (most cited first, 2 of 2)

- [Task-Agnostic Guided Feature Expansion for Class-Incremental Learning](https://mlsanity.com/paper/2503.00823/) (2025)
- [DetailSemNet: Elevating Signature Verification through Detail-Semantic Integration](https://mlsanity.com/paper/2511.16364/) (2025)

## Abstract (arXiv)

Learning new classes without forgetting is crucial for real-world applications for a classification model. Vision Transformers (ViT) recently achieve remarkable performance in Class Incremental Learning (CIL). Previous works mainly focus on block design and model expansion for ViTs. However, in this paper, we find that when the ViT is incrementally trained, the attention layers gradually lose concentration on local features. We call this interesting phenomenon as Locality Degradation in ViTs for CIL. Since the low-level local information is crucial to the transferability of the representation, it is beneficial to preserve the locality in attention layers. In this paper, we encourage the model to preserve more local information as the training procedure goes on and devise a Locality-Preserved Attention (LPA) layer to emphasize the importance of local features. Specifically, we incorporate the local information directly into the vanilla attention and control the initial gradients of the vanilla attention by weighting it with a small initial value. Extensive experiments show that the representations facilitated by LPA capture more low-level general information which is easier to transfer to follow-up tasks. The improved model gets consistently better performance on CIFAR100 and ImageNet100.