Catastrophic Forgetting

Ask about this topic

Overview

Catastrophic forgetting occurs when learning new tasks or adapting to shifting data degrades performance on earlier tasks, a central challenge for systems expected to learn over time. Common approaches include replaying examples, distilling prior outputs, regularizing important weights or gradients, and isolating updates with adapters or modular experts. Researchers study these methods in sequential vision and language benchmarks as well as foundation-model fine-tuning, multimodal systems, and robotics, measuring both new-task performance and retention. No single approach works best everywhere: stronger protection can limit adaptation, while memory, compute, and benchmark choices affect results; apparent forgetting may also reflect knowledge becoming inaccessible rather than erased.

Latest papers

CardsList

351 papers59 papers in the last four weeks, up 119% on the four weeks before. 0.3% of all new papers.

Mar 11, 2026

Representation Finetuning for Continual Learning

The world is inherently dynamic, and continual learning aims to enable models to adapt to ever-evolving data streams. While pre-trained models have shown powerful performance in continual learning, they still require finetuning to adapt effectively to downstream tasks. However, prevailing Parameter-Efficient Fine-Tuning (PEFT) methods operate through empirical, black-box optimization at the weight level. These approaches lack explicit control over representation drift, leading to sensitivity to domain shifts and catastrophic forgetting in continual learning scenarios. In this work, we introduce Continual Representation Learning (CoRe), a novel framework that for the first time shifts the finetuning paradigm from weight space to representation space. Unlike conventional methods, CoRe performs task-specific interventions within a low-rank linear subspace of hidden representations, adopting a learning process with explicit objectives, which ensures stability for past tasks while maintaining plasticity for new ones. By constraining updates to a low-rank subspace, CoRe achieves exceptional parameter efficiency. Extensive experiments across multiple continual learning benchmarks demonstrate that CoRe not only preserves parameter efficiency but also significantly outperforms existing state-of-the-art methods. Our work introduces representation finetuning as a new, more effective and interpretable paradigm for continual learning.

Haihua Luo, Xuming Ran, Tommi Kärkkäinen +4

Faculty of Information Technology, University of Jyväskylä, Jyväskylä, Finland · National University of Singapore, Singapore, Singapore · School of Computer Science and Technology, Dalian University of Technology, Dalian, China

Continual LearningFine-TuningParameter-Efficient Fine-TuningCatastrophic Forgetting
AskPDF
Mar 10, 2026

MSSR: Memory-Aware Adaptive Replay for Continual LLM Fine-Tuning

Continual fine-tuning of large language models (LLMs) is becoming increasingly crucial as these models are deployed in dynamic environments where tasks and data distributions evolve over time. While strong adaptability enables rapid acquisition of new knowledge, it also exposes LLMs to catastrophic forgetting, where previously learned skills degrade during sequential training. Existing replay-based strategies, such as fixed interleaved replay, accuracy-supervised, and loss-driven scheduling, remain limited: some depend on heuristic rules and provide only partial mitigation of forgetting, while others improve performance but incur substantial computational overhead. Motivated by retention dynamics under sequential fine-tuning, we propose Memory-Inspired Sampler and Scheduler Replay (MSSR), an experience replay framework that estimates sample-level memory strength and schedules rehearsal at adaptive intervals to mitigate catastrophic forgetting while maintaining fast adaptation. Extensive experiments across three backbone models and 11 sequential tasks show that MSSR consistently outperforms state-of-the-art replay baselines, with particularly strong gains on reasoning-intensive and multiple-choice benchmarks.

Yiyang Lu, Yu He, Jianlong Chen +1

The Chinese University of Hong Kong, Shenzhen · Nanjing University

Continual Learning for LLMsReplay-Based Continual LearningCatastrophic Forgetting
AskPDF
Mar 10, 2026

On Catastrophic Forgetting in Low-Rank Decomposition-Based Parameter-Efficient Fine-Tuning

Parameter-efficient fine-tuning (PEFT) based on low-rank decomposition, such as LoRA, has become a standard for adapting large pretrained models. However, its behavior in sequential learning -- specifically regarding catastrophic forgetting -- remains insufficiently understood. In this work, we present an empirical study showing that forgetting is strongly influenced by the geometry and parameterization of the update subspace. While methods that restrict updates to small, shared matrix subspaces often suffer from task interference, tensor-based decompositions (e.g., LoRETTA) mitigate forgetting by capturing richer structural information within ultra-compact budgets, and structurally aligned parameterizations (e.g., WeGeFT) preserve pretrained representations. Our findings highlight update subspace design as a key factor in continual learning and offer practical guidance for selecting efficient adaptation strategies in sequential settings.

Muhammad Ahmad, Jingjing Zheng, Yankai Cao

Department of Electrical and Computer Engineering, The University of British Columbia · Department of Mathematics, The University of British Columbia · Department of Chemical and Biological Engineering, The University of British Columbia

Parameter-Efficient Fine-TuningCatastrophic ForgettingLow-Rank AdaptationContinual Learning for LLMs
AskPDF
Mar 10, 2026

Guided Prompt Evolution for Vision-Language Models Adaptation

The adaptation of large-scale vision-language models (VLMs) to downstream tasks with limited labeled data remains a significant challenge. While parameter-efficient prompt learning methods offer a promising path, they often suffer from catastrophic forgetting of pre-trained knowledge. Toward addressing this limitation, our work is grounded in the insight that governing the evolutionary path of prompts is essential for forgetting-free adaptation. To this end, we propose EvoPrompt, a novel framework designed to explicitly steer the prompt trajectory for knowledge-preserving fine-tuning. Specifically, our approach employs a Modality-Shared Prompt Projector (MPP) to generate hierarchical prompts from a unified embedding space. Critically, an evolutionary training strategy decouples low-rank updates into directional and magnitude components, preserving early-learned semantic directions while only adapting their magnitude, thus enabling prompts to evolve without discarding foundational knowledge. Extensive experiments demonstrate that EvoPrompt achieves state-of-the-art performance in few-shot learning while robustly preserving the original zero-shot capabilities of pre-trained VLMs.

Enming Zhang, Jiayang Li, Yanlong Wang +3

Tsinghua Shenzhen International Graduate School, Tsinghua University · Sun Yat-sen University · Chinese University of Hong Kong, Shenzhen

VLM AdaptationFew-Shot LearningVisual Prompt LearningCatastrophic Forgetting
AskPDF
Mar 10, 2026

Causally Sufficient and Necessary Feature Expansion for Class-Incremental Learning

Current expansion-based methods for Class Incremental Learning (CIL) effectively mitigate catastrophic forgetting by freezing old features. However, such task-specific features learned from the new task may collide with the old features. From a causal perspective, spurious feature correlations are the main cause of this collision, manifesting in two scopes: (i) guided by empirical risk minimization (ERM), intra-task spurious correlations cause task-specific features to rely on shortcut features. These non-robust features are vulnerable to interference, inevitably drifting into the feature space of other tasks; (ii) inter-task spurious correlations induce semantic confusion between visually similar classes across tasks. To address this, we propose a Probability of Necessity and Sufficiency (PNS)-based regularization method to guide feature expansion in CIL. Specifically, we first extend the definition of PNS to expansion-based CIL, termed CPNS, which quantifies both the causal completeness of intra-task representations and the separability of inter-task representations. We then introduce a dual-scope counterfactual generator based on twin networks to ensure the measurement of CPNS, which simultaneously generates: (i) intra-task counterfactual features to minimize intra-task PNS risk and ensure causal completeness of task-specific features, and (ii) inter-task interfering features to minimize inter-task PNS risk, ensuring the separability of inter-task representations. Theoretical analyses confirm its reliability. The regularization is a plug-and-play method for expansion-based CIL to mitigate feature collision. Extensive experiments demonstrate the effectiveness of the proposed method.

Zhen Zhang, Jielei Chu, Jiangtao Hu +4

School of Computing and Artificial Intelligence, Southwest Jiaotong University, Chengdu 611756, China. · Engineering Research Center of Sustainable Urban Intelligent Transportation, Ministry of Education, Chengdu, Sichuan 611756, China · National Engineering Laboratory of Integrated Transportation Big Data Application Technology, Southwest Jiaotong University, Chengdu, Sichuan 611756, China. +2

Class-Incremental LearningCatastrophic Forgetting
AskPDF
Mar 9, 2026

Can You Hear, Localize, and Segment Continually? An Exemplar-Free Continual Learning Benchmark for Audio-Visual Segmentation

Audio-Visual Segmentation (AVS) aims to produce pixel-level masks of sound producing objects in videos, by jointly learning from audio and visual signals. However, real-world environments are inherently dynamic, causing audio and visual distributions to evolve over time, which challenge existing AVS systems that assume static training settings. To address this gap, we introduce the first exemplar-free continual learning benchmark for Audio-Visual Segmentation, comprising four learning protocols across single-source and multi-source AVS datasets. We further propose a strong baseline, ATLAS, which uses audio-guided pre-fusion conditioning to modulate visual feature channels via projected audio context before cross-modal attention. Finally, we mitigate catastrophic forgetting by introducing Low-Rank Anchoring (LRA), which stabilizes adapted weights based on loss sensitivity. Extensive experiments demonstrate competitive performance across diverse continual scenarios, establishing a foundation for lifelong audio-visual perception. Code is available at∗{}^{*}\footnote{Paper under review} - \hyperlink{https://gitlab.com/viper-purdue/atlas}{https://gitlab.com/viper-purdue/atlas} \keywords{Continual Learning \and Audio-Visual Segmentation \and Multi-Modal Learning}

Siddeshwar Raghavan, Gautham Vinod, Bruce Coburn +1

Purdue University, West Lafayette, IN 47906, USA

Audio-Visual UnderstandingContinual LearningCatastrophic Forgetting
AskPDF
Mar 9, 2026

Grow, Don't Overwrite: Fine-tuning Without Forgetting

Adapting pre-trained models to specialized tasks often leads to catastrophic forgetting, where new knowledge overwrites foundational capabilities. Existing methods either compromise performance on the new task or struggle to balance training stability with efficient reuse of pre-trained knowledge. We introduce a novel function-preserving expansion method that resolves this dilemma. Our technique expands model capacity by replicating pre-trained parameters within transformer submodules and applying a scaling correction that guarantees the expanded model is mathematically identical to the original at initialization, enabling stable training while exploiting existing knowledge. Empirically, our method eliminates the trade-off between plasticity and stability, matching the performance of full fine-tuning on downstream tasks without any degradation of the model's original capabilities. Furthermore, we demonstrate the modularity of our approach, showing that by selectively expanding a small subset of layers we can achieve the same performance as full fine-tuning at a fraction of the computational cost.

Dyah Adila, Hanna Mazzawi, Benoit Dherin +1

University of Wisconsin-Madison · Google Research

Catastrophic ForgettingParameter-Efficient Fine-Tuning
AskPDF
Mar 9, 2026

Don't Look Back in Anger: MAGIC Net for Streaming Continual Learning with Temporal Dependence

Concept drift, temporal dependence, and catastrophic forgetting represent major challenges when learning from data streams. While Streaming Machine Learning and Continual Learning (CL) address these issues separately, recent efforts in Streaming Continual Learning (SCL) aim to unify them. In this work, we introduce MAGIC Net, a novel SCL approach that integrates CL-inspired architectural strategies with recurrent neural networks to tame temporal dependence. MAGIC Net continuously learns, looks back at past knowledge by applying learnable masks over frozen weights, and expands its architecture when necessary. It performs all operations online, ensuring inference availability at all times. Experiments on synthetic and real-world streams show that it improves adaptation to new concepts, limits memory usage, and mitigates forgetting.

Federico Giannini, Sandro D'Andrea, Emanuele Della Valle

DEIB - Politecnico di Milano

Continual LearningOnline Continual LearningCatastrophic ForgettingConcept Drift
AskPDF
Mar 8, 2026

Context Channel Capacity: An Information-Theoretic Framework for Understanding Catastrophic Forgetting

Catastrophic forgetting remains a central challenge in continual learning (CL), yet lacks a unified information-theoretic explanation for why some architectures forget catastrophically while others do not. We introduce Context Channel Capacity (CctxC_\mathrm{ctx}), the mutual information between a CL architecture's context signal and its generated parameters, and prove that zero forgetting requires Cctx≥H(T)C_\mathrm{ctx} \geq H(T), where H(T)H(T) is the task identity entropy. We establish an Impossibility Triangle -- zero forgetting, online learning, and finite parameters cannot be simultaneously satisfied by sequential state-based learners -- and show that conditional regeneration architectures (HyperNetworks) bypass this triangle by redefining parameters as function values rather than states. We validate this framework across 8 CL methods on Split-MNIST (1,130+ experiments over 86 days, 4 seeds each), showing that CctxC_\mathrm{ctx} perfectly predicts forgetting behavior: methods with Cctx=0C_\mathrm{ctx} = 0 (NaiveSGD, EWC, SI, LwF, CFlow) exhibit catastrophic forgetting (6--97%), while methods with Cctx≈1C_\mathrm{ctx} \approx 1 (HyperNetwork) achieve zero forgetting (98.8% ACC). We further propose Wrong-Context Probing (P5), a practical diagnostic protocol for measuring CctxC_\mathrm{ctx}, and extend the framework to CIFAR-10 via a novel Gradient Context Encoder that closes the oracle gap from 23.3pp to 0.7pp. A systematic taxonomy of 15+ closed research directions -- including the Hebbian null result (frozen random features outperform learned features), CFlow's θ0θ_0-memorizer phenomenon, and the SNS_N symmetry barrier to column specialization -- provides the community with precisely diagnosed negative results. Our central design principle: architecture over algorithm -- the context pathway must be structurally unbypassable.

Ran Cheng

Catastrophic ForgettingOnline Continual Learning
AskPDF
Mar 8, 2026

UnSCAR: Universal, Scalable, Controllable, and Adaptable Image Restoration

Universal image restoration aims to recover clean images from arbitrary real-world degradations using a single inference model. Despite significant progress, existing all-in-one restoration networks do not scale to multiple degradations. As the number of degradations increases, training becomes unstable, models grow excessively large, and performance drops across both seen and unseen domains. In this work, we show that scaling universal restoration is fundamentally limited by interference across degradations during joint learning, leading to catastrophic task forgetting. To address this challenge, we introduce a unified inference pipeline with a multi-branch mixture-of-experts architecture that decomposes restoration knowledge across specialized task-adaptable experts. Our approach enables scalable learning (over sixteen degradations), adapts and generalizes robustly to unseen domains, and supports user-controllable restoration across degradations. Beyond achieving superior performance across benchmarks, this work establishes a new design paradigm for scalable and controllable universal image restoration.

Debabrata Mandal, Soumitri Chattopadhyay, Yujie Wang +2

University of North Carolina at Chapel Hill · University of California, San Diego

All-in-One Image RestorationCatastrophic Forgetting
AskPDF
Mar 6, 2026

Continual Adaptation for Pacific Indigenous Speech Recognition

Speech foundation models struggle with low-resource Pacific Indigenous languages because of severe data scarcity. Furthermore, full fine-tuning risks catastrophic forgetting. To address this gap, we present an empirical study adapting models to real-world Pacific datasets. We investigate the impact of data volume, adaptation strategies, and representational drift on speech foundation models for various Pacific languages. Additionally, we analyze a continual learning framework for sequential language acquisition. Empirical results across three distinct Pacific Indigenous languages demonstrate that adapting to these linguistically distant languages induces severe internal representational drift. Consequently, these models face a strict plasticity and stability dilemma. While LoRA adapts well initially, it suffers from catastrophic forgetting during sequential learning. Ultimately, this study highlights the urgent need for robust adaptation strategies tailored to underrepresented languages.

Yang Xiao, Aso Mahmudi, Nick Thieberger +3

The University of Melbourne · UNSW Sydney

Automatic Speech RecognitionLow-Resource Speech RecognitionSpeech Foundation ModelsRepresentation DriftCatastrophic Forgetting
AskPDF
Mar 5, 2026

Lifelong Language-Conditioned Robotic Manipulation Learning

Traditional language-conditioned manipulation agent sequential adaptation to new manipulation skills leads to catastrophic forgetting of old skills, limiting dynamic scene practical deployment. In this paper, we propose SkillsCrafter, a novel robotic manipulation framework designed to continually learn multiple skills while reducing catastrophic forgetting of old skills. Specifically, we propose a Manipulation Skills Adaptation to retain the old skills knowledge while inheriting the shared knowledge between new and old skills to facilitate learning of new skills. Meanwhile, we perform the singular value decomposition on the diverse skill instructions to obtain common skill semantic subspace projection matrices, thereby recording the essential semantic space of skills. To achieve forget-less and generalization manipulation, we propose a Skills Specialization Aggregation to compute inter-skills similarity in skill semantic subspaces, achieving aggregation of the previously learned skill knowledge for any new or unknown skill. Extensive experiments demonstrate the effectiveness and superiority of our proposed SkillsCrafter.

Xudong Wang, Zebin Han, Zhiyu Liu +5

State Key Laboratory of Robotics and Intelligent Systems, Shenyang Institute of Automation, Chinese Academy of Sciences · University of Chinese Academy of Sciences · North University of China +2

Language-Conditioned Robot ManipulationContinual Robot LearningCatastrophic ForgettingGeneralization in Robotic Manipulation
AskPDF
Mar 4, 2026

Why Do Neural Networks Forget: A Study of Collapse in Continual Learning

Catastrophic forgetting is a major problem in continual learning, and lots of approaches arise to reduce it. However, most of them are evaluated through task accuracy, which ignores the internal model structure. Recent research suggests that structural collapse leads to loss of plasticity, as evidenced by changes in effective rank (eRank). This indicates a link to forgetting, since the networks lose the ability to expand their feature space to learn new tasks, which forces the network to overwrite existing representations. Therefore, in this study, we investigate the correlation between forgetting and collapse through the measurement of both weight and activation eRank. To be more specific, we evaluated four architectures, including MLP, ConvGRU, ResNet-18, and Bi-ConvGRU, in the split MNIST and Split CIFAR-100 benchmarks. Those models are trained through the SGD, Learning-without-Forgetting (LwF), and Experience Replay (ER) strategies separately. The results demonstrate that forgetting and collapse are strongly related, and different continual learning strategies help models preserve both capacity and performance in different efficiency.

Yunqin Zhu, Jun Jin

Department of Electrical and Computer Engineering, University of Alberta, Edmonton, Canada

Catastrophic ForgettingIncremental LearningRepresentation CollapseNeural Network Training Dynamics
AskPDF
Mar 4, 2026

Pretrained Vision-Language-Action Models are Surprisingly Resistant to Forgetting in Continual Learning

Continual learning is a long-standing challenge in robot policy learning, where a policy must acquire new skills over time without catastrophically forgetting previously learned ones. While prior work has extensively studied continual learning in relatively small behavior cloning (BC) policy models trained from scratch, its behavior in modern large-scale pretrained Vision-Language-Action (VLA) models remains underexplored. In this work, we found that pretrained VLAs are remarkably resistant to forgetting compared with smaller policy models trained from scratch. Simple Experience Replay (ER) works surprisingly well on VLAs, sometimes achieving zero forgetting even with a small replay data size. Our analysis reveals that pretraining plays a critical role in downstream continual learning performance: large pretrained models mitigate forgetting with a small replay buffer size while maintaining strong forward learning capabilities. Furthermore, we found that VLAs can retain relevant knowledge from prior tasks despite performance degradation during learning new tasks. This knowledge retention enables rapid recovery of seemingly forgotten skills through finetuning. Together, these insights imply that large-scale pretraining fundamentally changes the dynamics of continual learning, enabling models to continually acquire new skills over time with simple replay. Code and more information can be found at https://continual-vlas.github.io/forget-me-not/

Huihan Liu, Changyeon Kim, Bo Liu +2

The University of Texas at Austin · KAIST · Microsoft Superintelligence.

Continual Robot LearningContinual Learning for VLA ModelsCatastrophic ForgettingReplay-Based Continual Learning
AskPDF
Mar 3, 2026

cPNN: Continuous Progressive Neural Networks for Evolving Streaming Time Series

Dealing with an unbounded data stream involves overcoming the assumption that data is identically distributed and independent. A data stream can, in fact, exhibit temporal dependencies (i.e., be a time series), and data can change distribution over time (concept drift). The two problems are deeply discussed, and existing solutions address them separately: a joint solution is absent. In addition, learning multiple concepts implies remembering the past (a.k.a. avoiding catastrophic forgetting in Neural Networks' terminology). This work proposes Continuous Progressive Neural Networks (cPNN), a solution that tames concept drifts, handles temporal dependencies, and bypasses catastrophic forgetting. cPNN is a continuous version of Progressive Neural Networks, a methodology for remembering old concepts and transferring past knowledge to fit the new concepts quickly. We base our method on Recurrent Neural Networks and exploit the Stochastic Gradient Descent applied to data streams with temporal dependencies. Results of an ablation study show a quick adaptation of cPNN to new concepts and robustness to drifts.

Federico Giannini, Giacomo Ziffer, Emanuele Della Valle

DEIB, Politecnico di Milano, Milano, Italy

Online Continual LearningConcept DriftCatastrophic Forgetting
AskPDF
Mar 2, 2026

Temporal Imbalance of Positive and Negative Supervision in Class-Incremental Learning

With the widespread adoption of deep learning in visual tasks, Class-Incremental Learning (CIL) has become an important paradigm for handling dynamically evolving data distributions. However, CIL faces the core challenge of catastrophic forgetting, often manifested as a prediction bias toward new classes. Existing methods mainly attribute this bias to intra-task class imbalance and focus on corrections at the classifier head. In this paper, we highlight an overlooked factor -- temporal imbalance -- as a key cause of this bias. Earlier classes receive stronger negative supervision toward the end of training, leading to asymmetric precision and recall. We establish a temporal supervision model, formally define temporal imbalance, and propose Temporal-Adjusted Loss (TAL), which uses a temporal decay kernel to construct a supervision strength vector and dynamically reweight the negative supervision in cross-entropy loss. Theoretical analysis shows that TAL degenerates to standard cross-entropy under balanced conditions and effectively mitigates prediction bias under imbalance. Extensive experiments demonstrate that TAL significantly reduces forgetting and improves performance on multiple CIL benchmarks, underscoring the importance of temporal modeling for stable long-term learning.

Jinge Ma, Fengqing Zhu

Purdue University, West Lafayette, Indiana, U.S.A.

Class-Incremental LearningCatastrophic ForgettingAdaptive Loss WeightingTemporal Distribution Shift
AskPDF
Mar 1, 2026

Principled Fast and Meta Knowledge Learners for Continual Reinforcement Learning

Inspired by the human learning and memory system, particularly the interplay between the hippocampus and cerebral cortex, this study proposes a dual-learner framework comprising a fast learner and a meta learner to address continual Reinforcement Learning~(RL) problems. These two learners are coupled to perform distinct yet complementary roles: the fast learner focuses on knowledge transfer, while the meta learner ensures knowledge integration. In contrast to traditional multi-task RL approaches that share knowledge through average return maximization, our meta learner incrementally integrates new experiences by explicitly minimizing catastrophic forgetting, thereby supporting efficient cumulative knowledge transfer for the fast learner. To facilitate rapid adaptation in new environments, we introduce an adaptive meta warm-up mechanism that selectively harnesses past knowledge. We conduct experiments in various pixel-based and continuous control benchmarks, revealing the superior performance of continual learning for our proposed dual-learner approach relative to baseline methods. The code is released in https://github.com/datake/FAME.

Ke Sun, Hongming Zhang, Jun Jin +4

University of Alberta · National Key Laboratory of Cognition and Decision Intelligence for Complex Systems, Institution of Automation, Chinese Academy of Sciences · Edmonton Research Center, Huawei Canada +2

Reinforcement LearningContinual RLMeta-Reinforcement LearningCatastrophic Forgetting
AskPDF
Feb 28, 2026

IDER: IDempotent Experience Replay for Reliable Continual Learning

Catastrophic forgetting, the tendency of neural networks to forget previously learned knowledge when learning new tasks, has been a major challenge in continual learning (CL). To tackle this challenge, CL methods have been proposed and shown to reduce forgetting. Furthermore, CL models deployed in mission-critical settings can benefit from uncertainty awareness by calibrating their predictions to reliably assess their confidences. However, existing uncertainty-aware continual learning methods suffer from high computational overhead and incompatibility with mainstream replay methods. To address this, we propose idempotent experience replay (IDER), a novel approach based on the idempotent property where repeated function applications yield the same output. Specifically, we first adapt the training loss to make model idempotent on current data streams. In addition, we introduce an idempotence distillation loss. We feed the output of the current model back into the old checkpoint and then minimize the distance between this reprocessed output and the original output of the current model. This yields a simple and effective new baseline for building reliable continual learners, which can be seamlessly integrated with other CL approaches. Extensive experiments on different CL benchmarks demonstrate that IDER consistently improves prediction reliability while simultaneously boosting accuracy and reducing forgetting. Our results suggest the potential of idempotence as a promising principle for deploying efficient and trustworthy continual learning systems in real-world applications.Our code is available at https://github.com/YutingLi0606/Idempotent-Continual-Learning.

Zhanwang Liu, Yuting Li, Haoyuan Gao +4

School of Computer Science, Shanghai Jiao Tong University · State Key Laboratory of General Artificial Intelligence, BIGAI · Lehigh University +1

Continual LearningCatastrophic ForgettingReplay-Based Continual Learning
AskPDF
Feb 28, 2026

Decoupling Stability and Plasticity for Multi-Modal Test-Time Adaptation

Adapting pretrained multi-modal models to evolving test-time distributions, known as multi-modal test-time adaptation, presents a significant challenge. Existing methods frequently encounter negative transfer in the unbiased modality and catastrophic forgetting in the biased modality. To address these challenges, we propose Decoupling Adaptation for Stability and Plasticity (DASP), a novel diagnose-then-mitigate framework. Our analysis reveals a critical discrepancy within the unified latent space: the biased modality exhibits substantially higher interdimensional redundancy (i.e., strong correlations across feature dimensions) compared to the unbiased modality. Leveraging this insight, DASP identifies the biased modality and implements an asymmetric adaptation strategy. This strategy employs a decoupled architecture where each modality-specific adapter is divided into stable and plastic components. The asymmetric mechanism works as follows: for the biased modality, which requires plasticity, the plastic component is activated and updated to capture domain-specific information, while the stable component remains fixed. Conversely, for the unbiased modality, which requires stability, the plastic component is bypassed, and the stable component is updated using KL regularization to prevent negative transfer. This asymmetric design enables the model to adapt flexibly to new domains while preserving generalizable knowledge. Comprehensive evaluations on diverse multi-modal benchmarks demonstrate that DASP significantly outperforms state-of-the-art methods.

Yongbo He, Zirun Guo, Tao Jin

Zhejiang University

Test-Time AdaptationTest-Time Adaptation of VLMsContinual Test-Time AdaptationCatastrophic Forgetting
AskPDF
Feb 27, 2026

Quantifying Catastrophic Forgetting in IoT Intrusion Detection Systems

Distribution shifts in attack patterns within RPL-based IoT networks pose a critical threat to the reliability and security of large-scale connected systems. Intrusion Detection Systems (IDS) trained on static datasets often fail to generalize to unseen threats and suffer from catastrophic forgetting when updated with new attacks. Ensuring continual adaptability of IDS is therefore essential for maintaining robust IoT network defence. In this focused study, we formulate intrusion detection as a domain continual learning problem and propose a method-agnostic IDS framework that can integrate diverse continual learning strategies. We systematically benchmark five representative approaches across multiple domain-ordering sequences using a comprehensive multi-attack dataset comprising 48 domains. Results show that continual learning mitigates catastrophic forgetting while maintaining a balance between plasticity, stability, and efficiency, a crucial aspect for resource-constrained IoT environments. Among the methods, Replay-based approaches achieve the best overall performance, while Synaptic Intelligence (SI) delivers near-zero forgetting with high training efficiency, demonstrating strong potential for stable and sustainable IDS deployment in dynamic IoT networks.

Sourasekhar Banerjee, David Bergqvist, Salman Toor +2

Uppsala University, Department of Information Technology, Sweden · Scaleout Systems AB, Sweden · Ericsson Research, Research Area AI, Sweden

IoT Intrusion DetectionCatastrophic ForgettingDomain-Incremental LearningReplay-Based Continual Learning
AskPDF
Feb 26, 2026

Fine-Tuning Without Forgetting In-Context Learning: A Theoretical Analysis of Linear Attention Models

Transformer-based large language models exhibit in-context learning, enabling adaptation to downstream tasks via few-shot prompting with demonstrations. In practice, such models are often fine-tuned to improve zero-shot performance on downstream tasks, allowing them to solve tasks without examples and thereby reducing inference costs. However, fine-tuning can degrade in-context learning, limiting the performance of fine-tuned models on tasks not seen during fine-tuning. Using linear attention models, we provide a theoretical analysis that characterizes how fine-tuning objectives modify attention parameters and identifies conditions under which this leads to degraded few-shot performance. We show that fine-tuning all attention parameters can harm in-context learning, whereas restricting updates to the value matrix improves zero-shot performance while preserving in-context learning. We further show that incorporating an auxiliary few-shot loss enhances in-context learning primarily on the target task, at the expense of degraded in-context learning ability on tasks not seen during fine-tuning. We provide empirical evidence from synthetic and real-world datasets consistent with the qualitative predictions of our theory.

Chungpa Lee, Jy-yong Sohn, Kangwook Lee

Yonsei University, Korea · University of Wisconsin–Madison, USA · KRAFTON, Korea +1

LLM Fine-TuningIn-Context LearningCatastrophic Forgetting
AskPDF
Feb 26, 2026

φφ-DPO: Fairness Direct Preference Optimization Approach to Continual Learning in Large Multimodal Models

Fairness in Continual Learning for Large Multimodal Models (LMMs) is an emerging yet underexplored challenge, particularly in the presence of imbalanced data distributions that can lead to biased model updates and suboptimal performance across tasks. While recent continual learning studies have made progress in addressing catastrophic forgetting, the problem of fairness caused the imbalanced data remains largely underexplored. This paper presents a novel Fairness Direct Preference Optimization (FaiDPO or φφ-DPO) framework for continual learning in LMMs. In particular, we first propose a new continual learning paradigm based on Direct Preference Optimization (DPO) to mitigate catastrophic forgetting by aligning learning with pairwise preference signals. Then, we identify the limitations of conventional DPO in imbalanced data and present a new φφ-DPO loss that explicitly addresses distributional biases. We provide a comprehensive theoretical analysis demonstrating that our approach addresses both forgetting and data imbalance. Additionally, to enable φφ-DPO-based continual learning, we construct pairwise preference annotations for existing benchmarks in the context of continual learning. Extensive experiments and ablation studies show the proposed φφ-DPO achieves State-of-the-Art performance across multiple benchmarks, outperforming prior continual learning methods of LMMs.

Thanh-Dat Truong, Huu-Thien Tran, Jackson Cothren +2

CVIU Lab, University of Arkansas, USA · Dep. of Geosciences, University of Arkansas, USA · Carnegie Mellon University, USA

Continual LearningDirect Preference OptimizationCatastrophic ForgettingClass-Imbalanced Learning
AskPDF
Feb 24, 2026

Exploring the Impact of Parameter Update Magnitude on Forgetting and Generalization of Continual Learning

The magnitude of parameter updates are considered a key factor in continual learning. However, most existing studies focus on designing diverse update strategies, while a theoretical understanding of the underlying mechanisms remains limited. Therefore, we characterize model's forgetting from the perspective of parameter update magnitude and formalize it as knowledge degradation induced by task-specific drift in the parameter space, which has not been fully captured in previous studies due to their assumption of a unified parameter space. By deriving the optimal parameter update magnitude that minimizes forgetting, we unify two representative update paradigms, frozen training and initialized training, within an optimization framework for constrained parameter updates. Our theoretical results further reveals that sequence tasks with small parameter distances exhibit better generalization and less forgetting under frozen training rather than initialized training. These theoretical insights inspire a novel hybrid parameter update strategy that adaptively adjusts update magnitude based on gradient directions. Experiments on deep neural networks demonstrate that this hybrid approach outperforms standard training strategies, providing new theoretical perspectives and practical inspiration for designing efficient and scalable continual learning algorithms.

JinLi He, Liang Bai, Xian Yang

Institute of Intelligent Information Processing, Shanxi University, Taiyuan, 030006, China · Alliance Manchester Business School, The University of Manchester, Manchester, M13 9PL, UK

Continual LearningCatastrophic Forgetting
AskPDF
Feb 24, 2026

Understanding the Role of Rehearsal Scale in Continual Learning under Varying Model Capacities

Rehearsal is one of the key techniques for mitigating catastrophic forgetting and has been widely adopted in continual learning algorithms due to its simplicity and practicality. However, the theoretical understanding of how rehearsal scale influences learning dynamics remains limited. To address this gap, we formulate rehearsal-based continual learning as a multidimensional effectiveness-driven iterative optimization problem, providing a unified characterization across diverse performance metrics. Within this framework, we derive a closed-form analysis of adaptability, memorability, and generalization from the perspective of rehearsal scale. Our results uncover several intriguing and counterintuitive findings. First, rehearsal can impair model's adaptability, in sharp contrast to its traditionally recognized benefits. Second, increasing the rehearsal scale does not necessarily improve memory retention. When tasks are similar and noise levels are low, the memory error exhibits a diminishing lower bound. Finally, we validate these insights through numerical simulations and extended analyses on deep neural networks across multiple real-world datasets, revealing statistical patterns of rehearsal mechanisms in continual learning.

JinLi He, Liang Bai, Xian Yang

Institute of Intelligent Information Processing, Shanxi University, Taiyuan, 030006, China · Alliance Manchester Business School, The University of Manchester, Manchester, M13 9PL, UK

Continual LearningReplay-Based Continual LearningCatastrophic ForgettingNeural Network Generalization
AskPDF
Feb 21, 2026

GONE: Structural Knowledge Unlearning via Neighborhood-Expanded Distribution Shaping

Unlearning knowledge is a pressing and challenging task in Large Language Models (LLMs) because of their unprecedented capability to memorize and digest training data at scale, raising more significant issues regarding safety, privacy, and intellectual property. However, existing works, including parameter editing, fine-tuning, and distillation-based methods, are all focused on flat sentence-level data but overlook the relational, multi-hop, and reasoned knowledge in naturally structured data. In response to this gap, this paper introduces Graph Oblivion and Node Erasure (GONE), a benchmark for evaluating knowledge unlearning over structured knowledge graph (KG) facts in LLMs.This KG-based benchmark enables the disentanglement of three effects of unlearning: direct fact removal, reasoning-based leakage, and catastrophic forgetting. In addition, Neighborhood-Expanded Distribution Shaping (NEDS), a novel unlearning framework, is designed to leverage graph connectivity and identify anchor-correlated neighbors, thereby enforcing a precise semantic separation between the forgotten fact and its semantic neighborhood. Evaluations on LLaMA-3-8B and Mistral-7B across multiple knowledge editing and unlearning methods showcase NEDS's superior performance (1.000 on unlearning efficacy and 0.839 on locality) on GONE and other benchmarks. The dataset is available at https://huggingface.co/datasets/GONE-Anonymous/GONE.

Chahana Dahal, Ashutosh Balasubramaniam, Zuobin Xiong

University of Nevada, Las Vegas · Indian Institute of Technology Guwahati

Machine UnlearningLLM UnlearningCatastrophic Forgetting
AskPDF
Feb 20, 2026

Non-Interfering Weight Fields: Treating Model Parameters as a Continuously Extensible Function

Large language models store all learned knowledge in a single, fixed weight vector. Teaching a model new capabilities requires modifying those same weights, inevitably degrading previously acquired knowledge. This fundamental limitation, known as catastrophic forgetting, has resisted principled solutions for decades. Existing approaches treat weights as immutable artifacts that must be protected through techniques like regularization heuristics, replay buffers, or isolated adapter modules. The problem is none of these provide a structural guarantee against forgetting. In this work, we propose Non-Interfering Weight Fields (NIWF), a framework that replaces the fixed weight paradigm with a learned function that generates weight configurations on demand from a continuous capability coordinate space. After training on a task, we commit the occupied coordinate region by snapshotting the fields outputs on anchor points to enforce a functional lock during all future training. We validate NIWF on sequential instructionfollowing and code generation tasks using Mistral-7B, demonstrating zero forgetting on committed tasks with competitive perplexity on new tasks. The framework introduces the notion of software-like versioning for neural network intelligence, where capabilities can be committed, extended, composed, and rolled back without retraining.

Sarim Chaudhry

Purdue University

Continual Learning for LLMsCatastrophic ForgettingHypernetworks
AskPDF
Feb 20, 2026

Audio-Visual Continual Test-Time Adaptation without Forgetting

Audio-visual continual test-time adaptation involves continually adapting a source audio-visual model at test-time, to unlabeled non-stationary domains, where either or both modalities can be distributionally shifted, which hampers online cross-modal learning and eventually leads to poor accuracy. While previous works have tackled this problem, we find that SOTA methods suffer from catastrophic forgetting where the model's performance drops well below even the source model due to continual parameter updates at test-time. In this work, we first show that adapting only the modality fusion layer to a target domain not only improves performance on that domain but can also enhance performance on subsequent domains. Based on this strong cross-task transferability of the fusion layer's parameters, we propose a method, AVReCAP\texttt{AVReCAP}, that improves test-time performance of the models without access to any source data. Our approach works by using a selective parameter retrieval mechanism that dynamically retrieves the best fusion layer parameters from a buffer using only a small batch of test data. These parameters are then integrated into the model, adapted to the current test distribution, and saved back for future use. Extensive experiments on benchmark datasets involving unimodal and bimodal corruptions show our proposed AVReCAP\texttt{AVReCAP} significantly outperforms existing methods while minimizing catastrophic forgetting.

Sarthak Kumar Maharana, Akshay Mehra, Bhavya Ramakrishna +2

The University of Texas at Dallas, Richardson, TX 75080, USA · Dolby Laboratories, Inc., San Francisco, CA 94103, USA

Cross-Modal LearningSource-Free Domain AdaptationAudio-Visual UnderstandingContinual Test-Time AdaptationCatastrophic Forgetting
AskPDF
Feb 19, 2026

Catastrophic Forgetting Resilient One-Shot Incremental Federated Learning

Modern big-data systems generate massive, heterogeneous, and geographically dispersed streams that are large-scale and privacy-sensitive, making centralization challenging. While federated learning (FL) provides a privacy-enhancing training mechanism, it assumes a static data flow and learns a collaborative model over multiple rounds, making learning with incremental data challenging in limited-communication scenarios. This paper presents One-Shot Incremental Federated Learning (OSI-FL), the first FL framework that addresses the dual challenges of communication overhead and catastrophic forgetting. OSI-FL communicates category-specific embeddings, devised by a frozen vision-language model (VLM) from each client in a single communication round, which a pre-trained diffusion model at the server uses to synthesize new data similar to the client's data distribution. The synthesized samples are used on the server for training. However, two challenges still persist: i) tasks arriving incrementally need to retrain the global model, and ii) as future tasks arrive, retraining the model introduces catastrophic forgetting. To this end, we augment training with Selective Sample Retention (SSR), which identifies and retains the top-p most informative samples per category and task pair based on sample loss. SSR bounds forgetting by ensuring that representative retained samples are incorporated into training in further iterations. The experimental results indicate that OSI-FL outperforms baselines, including traditional and one-shot FL approaches, in both class-incremental and domain-incremental scenarios across three benchmark datasets.

Obaidullah Zaland, Zulfiqar Ahmad Khan, Monowar Bhuyan

Department of Computing Science, Umeå University, Umeå, SE-90187, Sweden

Federated LearningFederated Continual LearningCatastrophic ForgettingClass-Incremental LearningCommunication-Efficient Distributed TrainingDomain-Incremental Learning
AskPDF
Feb 19, 2026

CapTrack: Multifaceted Evaluation of Forgetting in LLM Post-Training

Large language model (LLM) post-training enhances latent skills, unlocks value alignment, improves performance, and enables domain adaptation. Unfortunately, post-training is known to induce forgetting, especially in the ubiquitous use-case of leveraging third-party pre-trained models, which is typically understood as a loss of parametric or factual knowledge. We argue that this accuracy-centric view is insufficient for modern foundation models and instead define forgetting as systematic model drift that degrades behavior and user experience. In this context, we introduce CapTrack, a capability-centric framework for analyzing forgetting in LLMs that combines a behavioral taxonomy with an evaluation suite centered on capability-specific metrics. Using CapTrack, we conduct a large-scale empirical study across post-training algorithms, domains, and model families, including models up to 80B parameters. We find that forgetting extends beyond parametric knowledge, with pronounced drift in robustness and default behaviors. Instruction fine-tuning induces the strongest relative drift, while preference optimization is more conservative and can partially recover lost capabilities. Differences across model families persist, and no universal mitigation emerges.

Lukas Thede, Stefan Winzeck, Zeynep Akata +1

Thomson Reuters Foundational Research, 5 Canada Square, London · Tübingen AI Center, University of Tübingen · Helmholtz Munich +2

LLM Post-TrainingCatastrophic ForgettingLLM EvaluationLanguage Model Robustness
AskPDF
Feb 17, 2026

Updating Parametric Knowledge with Context Distillation Retains Post-Training Capabilities

Post-training endows pretrained LLMs with a variety of desirable skills, including instruction-following, reasoning, and others. However, these post-trained LLMs only encode knowledge up to a cut-off date, necessitating continual adaptation. Unfortunately, existing solutions cannot simultaneously learn new knowledge from an adaptation document corpora and mitigate the forgetting of earlier learned capabilities. To address this, we introduce Distillation via Split Contexts (DiSC), a simple context-distillation based approach for continual knowledge adaptation. \methodname~derives student and teacher distributions by conditioning on distinct segments of the training example and minimizes the KL divergence between the shared tokens. This allows us to efficiently apply context-distillation without requiring explicit generation steps during training. We run experiments on four post-trained models and two adaptation domains. Compared to prior finetuning and distillation methods for continual adaptation, DiSC consistently reports the best trade-off between learning new knowledge and mitigating forgetting of previously learned skills like instruction-following, reasoning, and factual knowledge.

Shankar Padmanabhan, Mustafa Omer Gul, Tanya Goyal

Department of Computer Science, Cornell University.

Continual Learning for LLMsContext DistillationCatastrophic Forgetting
AskPDF