Memory-Modular Classification: Learning to Generalize with Memory Replacement

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Kang, Dahyun, Iscen, Ahmet, Jo, Eunchan, Choi, Sua, Cho, Minsu, Schmid, Cordelia
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909571737452544
author Kang, Dahyun
Iscen, Ahmet
Jo, Eunchan
Choi, Sua
Cho, Minsu
Schmid, Cordelia
author_facet Kang, Dahyun
Iscen, Ahmet
Jo, Eunchan
Choi, Sua
Cho, Minsu
Schmid, Cordelia
contents We propose a novel memory-modular learner for image classification that separates knowledge memorization from reasoning. Our model enables effective generalization to new classes by simply replacing the memory contents, without the need for model retraining. Unlike traditional models that encode both world knowledge and task-specific skills into their weights during training, our model stores knowledge in the external memory of web-crawled image and text data. At inference time, the model dynamically selects relevant content from the memory based on the input image, allowing it to adapt to arbitrary classes by simply replacing the memory contents. The key differentiator that our learner meta-learns to perform classification tasks with noisy web data from unseen classes, resulting in robust performance across various classification scenarios. Experimental results demonstrate the promising performance and versatility of our approach in handling diverse classification tasks, including zero-shot/few-shot classification of unseen classes, fine-grained classification, and class-incremental classification.
format Preprint
id arxiv_https___arxiv_org_abs_2504_06021
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Memory-Modular Classification: Learning to Generalize with Memory Replacement
Kang, Dahyun
Iscen, Ahmet
Jo, Eunchan
Choi, Sua
Cho, Minsu
Schmid, Cordelia
Computer Vision and Pattern Recognition
We propose a novel memory-modular learner for image classification that separates knowledge memorization from reasoning. Our model enables effective generalization to new classes by simply replacing the memory contents, without the need for model retraining. Unlike traditional models that encode both world knowledge and task-specific skills into their weights during training, our model stores knowledge in the external memory of web-crawled image and text data. At inference time, the model dynamically selects relevant content from the memory based on the input image, allowing it to adapt to arbitrary classes by simply replacing the memory contents. The key differentiator that our learner meta-learns to perform classification tasks with noisy web data from unseen classes, resulting in robust performance across various classification scenarios. Experimental results demonstrate the promising performance and versatility of our approach in handling diverse classification tasks, including zero-shot/few-shot classification of unseen classes, fine-grained classification, and class-incremental classification.
title Memory-Modular Classification: Learning to Generalize with Memory Replacement
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2504.06021