Dual-LoRA and Quality-Enhanced Pseudo Replay for Multimodal Continual Food Learning

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Wu, Xinlan, Zhu, Bin, Han, Feng, Jiao, Pengkun, Chen, Jingjing
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866908659485769728
author Wu, Xinlan
Zhu, Bin
Han, Feng
Jiao, Pengkun
Chen, Jingjing
author_facet Wu, Xinlan
Zhu, Bin
Han, Feng
Jiao, Pengkun
Chen, Jingjing
contents Food analysis has become increasingly critical for health-related tasks such as personalized nutrition and chronic disease prevention. However, existing large multimodal models (LMMs) in food analysis suffer from catastrophic forgetting when learning new tasks, requiring costly retraining from scratch. To address this, we propose a novel continual learning framework for multimodal food learning, integrating a Dual-LoRA architecture with Quality-Enhanced Pseudo Replay. We introduce two complementary low-rank adapters for each task: a specialized LoRA that learns task-specific knowledge with orthogonal constraints to previous tasks' subspaces, and a cooperative LoRA that consolidates shared knowledge across tasks via pseudo replay. To improve the reliability of replay data, our Quality-Enhanced Pseudo Replay strategy leverages self-consistency and semantic similarity to reduce hallucinations in generated samples. Experiments on the comprehensive Uni-Food dataset show superior performance in mitigating forgetting, representing the first effective continual learning approach for complex food tasks.
format Preprint
id arxiv_https___arxiv_org_abs_2511_13351
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Dual-LoRA and Quality-Enhanced Pseudo Replay for Multimodal Continual Food Learning
Wu, Xinlan
Zhu, Bin
Han, Feng
Jiao, Pengkun
Chen, Jingjing
Machine Learning
Artificial Intelligence
Food analysis has become increasingly critical for health-related tasks such as personalized nutrition and chronic disease prevention. However, existing large multimodal models (LMMs) in food analysis suffer from catastrophic forgetting when learning new tasks, requiring costly retraining from scratch. To address this, we propose a novel continual learning framework for multimodal food learning, integrating a Dual-LoRA architecture with Quality-Enhanced Pseudo Replay. We introduce two complementary low-rank adapters for each task: a specialized LoRA that learns task-specific knowledge with orthogonal constraints to previous tasks' subspaces, and a cooperative LoRA that consolidates shared knowledge across tasks via pseudo replay. To improve the reliability of replay data, our Quality-Enhanced Pseudo Replay strategy leverages self-consistency and semantic similarity to reduce hallucinations in generated samples. Experiments on the comprehensive Uni-Food dataset show superior performance in mitigating forgetting, representing the first effective continual learning approach for complex food tasks.
title Dual-LoRA and Quality-Enhanced Pseudo Replay for Multimodal Continual Food Learning
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2511.13351