InfLoRA: Interference-Free Low-Rank Adaptation for Continual Learning

2024/03/30 by Yan-Shuo Liang, Liang, Yan-Shuo, Wu-Jun Li +1 · 63 citations
Computer Science · Engineering · #Artificial Intelligence (cs.AI) #Computer Vision and Pattern Recognition (cs.CV) #Domain Adaptation and Few-Shot Learning #FOS: Computer and information sciences #Indoor and Outdoor Localization Technologies #Machine Learning (cs.LG) #Speech and Audio Processing

paper · pdf · doi:10.48550/arxiv.2404.00228

openalex publication_date 2024/03/30 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

Continual learning requires the model to learn multiple tasks sequentially. In continual learning, the model should possess the ability to maintain its performance on old tasks (stability) and the ability to adapt to new tasks continuously (plasticity). Recently, parameter-efficient fine-tuning (PEFT), which involves freezing a pre-trained model and injecting a small number of learnable parameters to adapt to downstream tasks, has gained increasing popularity in continual learning. Although existing continual learning methods based on PEFT have demonstrated superior performance compared to those not based on PEFT, most of them do not consider how to eliminate the interference of the new task on the old tasks, which inhibits the model from making a good trade-off between stability and plasticity. In this work, we propose a new PEFT method, called interference-free low-rank adaptation (InfLoRA), for continual learning. InfLoRA injects a small number of parameters to reparameterize the pre-trained weights and shows that fine-tuning these injected parameters is equivalent to fine-tuning the pre-trained weights within a subspace. Furthermore, InfLoRA designs this subspace to eliminate the interference of the new task on the old tasks, making a good trade-off between stability and plasticity. Experimental results show that InfLoRA outperforms existing state-of-the-art continual learning methods on multiple datasets.

Cited by

• Merge before Forget: A Single LoRA Continual Learning via Continual Merging
• Latent-LoRA: Compact Latent-Space Adapters with Gradient-Free Routing for Continual Learning
• LibContinual: A Comprehensive Library towards Realistic Continual Learning
• Efficient Continual Learning in Neural Machine Translation: A Low-Rank Adaptation Approach
• Resolving Conflicts in Lifelong Learning via Aligning Updates in Subspaces
• InTAct: Interval-based Task Activation Consolidation for Continual Learning
• AnaCP: Toward Upper-Bound Continual Learning via Analytic Contrastive Projection
• Teaching Prompts to Coordinate: Hierarchical Layer-Grouped Prompt Tuning for Continual Learning
• Multi-Modal Continual Learning via Cross-Modality Adapters and Representation Alignment with Knowledge Preservation
• Model Inversion with Layer-Specific Modeling and Alignment for Data-Free Continual Learning
• Continual Low-Rank Adapters for LLM-based Generative Recommender Systems
• PLAN: Proactive Low-Rank Allocation for Continual Learning
• Memory-Free Continual Learning with Null Space Adaptation for Zero-Shot Vision-Language Models
• More Than Memory Savings: Zeroth-Order Optimization Mitigates Forgetting in Continual Learning
• Fly-CL: A Fly-Inspired Framework for Enhancing Efficient Decorrelation and Reduced Training Time in Pre-trained Model-based Continual Representation Learning
• DOLFIN: Balancing Stability and Plasticity in Federated Continual Learning
• Deep Generative Continual Learning using Functional LoRA: FunLoRA
• OrthAlign: Orthogonal Subspace Decomposition for Non-Interfering Multi-Objective Alignment
• LEAF: A Robust Expert-Based Framework for Few-Shot Continual Event Detection
• LANCE: Low Rank Activation Compression for Efficient On-Device Continual Learning
• Task-Agnostic Federated Continual Learning via Replay-Free Gradient Projection
• Null-Space Filtering for Data-Free Continual Model Merging: Preserving Transparency, Promoting Fidelity
• Adaptive Model Ensemble for Continual Learning
• Policy Compatible Skill Incremental Learning via Lazy Learning Interface
• Continual Learning with Vision-Language Models via Semantic-Geometry Preservation
• DevFD: Developmental Face Forgery Detection by Learning Shared and Orthogonal LoRA Subspaces
• HAM: Hierarchical Adapter Merging for Scalable Continual Learning
• Modular Embedding Recomposition for Incremental Learning
• Contrastive Regularization over LoRA for Multimodal Biomedical Image Incremental Learning
• Sculpting Margin Penalty: Intra-Task Adapter Merging and Classifier Calibration for Few-Shot Class-Incremental Learning
• Distribution-aware Knowledge Unification and Association for Non-exemplar Lifelong Person Re-identification
• Harnessing Textual Semantic Priors for Knowledge Transfer and Refinement in CLIP-Driven Continual Learning
• Instruction-Grounded Visual Projectors for Continual Learning of Generative Vision-Language Models
• RainbowPrompt: Diversity-Enhanced Prompt-Evolving for Continual Learning
• PROL : Rehearsal Free Continual Learning in Streaming Data via Prompt Online Learning
• CKAA: Cross-subspace Knowledge Alignment and Aggregation for Robust Continual Learning
• SAS: Simulated Attention Score
• Continual Multiple Instance Learning with Enhanced Localization for Histopathological Whole Slide Image Analysis
• DoMIX: An Efficient Framework for Exploiting Domain Knowledge in Fine-Tuning
• Little by Little: Continual Learning via Incremental Mixture of Rank-1 Associative Memory Experts
• Consistency-Driven Calibration and Matching for Few-Shot Class-Incremental Learning
• EKPC: Elastic Knowledge Preservation and Compensation for Class-Incremental Learning
• Componential Prompt-Knowledge Alignment for Domain Incremental Learning
• LoRA as a Flexible Framework for Securing Large Vision Systems
• Flashbacks to Harmonize Stability and Plasticity in Continual Learning
• CL-LoRA: Continual Low-Rank Adaptation for Rehearsal-Free Class-Incremental Learning
• SplitLoRA: Balancing Stability and Plasticity in Continual Learning Through Gradient Space Splitting
• Continual Learning Beyond Experience Rehearsal and Full Model Surrogates
• Train with Perturbation, Infer after Merging: A Two-Stage Framework for Continual Learning
• Continual Learning on CLIP via Incremental Prompt Tuning with Intrinsic Textual Anchors
• Exemplar-Free Continual Learning for State Space Models
• Dynamic Dual Buffer with Divide-and-Conquer Strategy for Online Continual Learning
• Gated Integration of Low-Rank Adaptation for Continual Learning of Large Language Models
• StPR: Spatiotemporal Preservation and Routing for Exemplar-Free Video Class-Incremental Learning
• Task-agnostic Low-rank Residual Adaptation for Efficient Federated Continual Fine-Tuning
• Parameter Efficient Continual Learning with Dynamic Low-Rank Adaptation
• Continuous Subspace Optimization for Continual Learning
• Beyond CLIP Generalization: Against Forward&Backward Forgetting Adapter for Continual Learning of Vision-Language Models
• Shared LoRA Subspaces for almost Strict Continual Learning
• TailLoR: Protecting Principal Components in Parameter-Efficient Continual Learning
• PEFT-Arena: Understanding Parameter-Efficient Finetuning from a Stability-Plasticity Perspective
• LoRA-Based Continual Learning with Constraints on Critical Parameter Changes
• CP-MoE: Consistency-Preserving Mixture-of-Experts for Continual Learning
• Adaptive Additive Parameter Updates of Vision Transformers for Few-Shot Continual Learning

Related

• PLAN: Proactive Low-Rank Allocation for Continual Learning
• SAFE: Slow and Fast Parameter-Efficient Tuning for Continual Learning with Pre-Trained Models
• Parameter-Efficient Continual Fine-Tuning: A Survey
• SplitLoRA: Balancing Stability and Plasticity in Continual Learning Through Gradient Space Splitting
• Weighted Ensemble Models Are Strong Continual Learners