Repurposing Adversarial Perturbations for Continual Learning: From Defense to Active Alignment

• URL: http://arxiv.org/abs/2606.02322v1
• Date: Mon, 01 Jun 2026 14:35:59 GMT
• Title: Repurposing Adversarial Perturbations for Continual Learning: From Defense to Active Alignment
• Abstract summary: AdvCL combines three plug-in modules: Intra-Smooth promotes local smoothness via small adversarial perturbations; Proto-Clip uses similarity clipping to prevent excessive alignment to current task prototype; and Inter-Align applies directional alignment toward previous task prototype to reduce representational gaps.<n>Experiments show consistent gains in both standard performance and robustness, with lower forgetting and stronger transfer.
• Score: 20.37950128370149
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: In dynamic environments, large language models need to keep adapting to new tasks, but continual learning often suffers from forgetting, limited transfer, and vulnerability to adversarial perturbations. To address this, we present AdvCL, which repurposes adversarial perturbations as a geometric control signal for stable continual adaptation. AdvCL combines three plug-in modules: Intra-Smooth promotes local smoothness via small adversarial perturbations; Proto-Clip uses similarity clipping to prevent excessive alignment to current task prototype; and Inter-Align applies directional alignment toward previous task prototype to reduce representational gaps. Experiments show consistent gains in both standard performance and robustness, with lower forgetting and stronger transfer. We further analyze key mechanisms by quantifying the sensitivity of Intra-Smooth to perturbation settings and the effect of Inter-Align on task similarity and geometric distance. In summary, the modules provide complementary gains when combined, and each can also be integrated individually into diverse CL paradigms, including replay, regularization, and dynamic architectures, thereby offering a geometric control mechanism for continual learning.

Related papers

• Isotropic Embedding Perturbations for Robust Vision Language Encoders [73.92626712781859]
  Aether is a simple plug-in method that applies diffusion-style random perturbations in the embedding space via controlled alpha-mixing.<n>Inspired by feature-space perturbations in language models and image degradation in generative pretraining, Aether induces mild yet effective perturbations.<n>Aether delivers consistent gains over the advanced recipe combining CutMix, Mixup, DropPath, and RandAug.
  arXiv  Detail & Related papers  (2026-09-09T15:08:05Z)
• The Quadrilateral Loss: Additivity as a Measurable Behavior of Dense Neural Networks [0.0]
  We introduce the quadrilateral loss, a differentiable penalty that treats additivity as a measurable behavior instead.<n>We compare routes to exact additivity, spanning structural masks, behavioral penalties, weight decay, backfitting, the shared-section model, and bagged boosted stumps.
  arXiv  Detail & Related papers  (2026-07-22T14:23:37Z)
• A Task-Space Receding Horizon Controller for Fast Collision Avoidance [0.08594140167290099]
  Real-time collision avoidance for robotic manipulators requires fast reactions to unexpected obstacle motion.<n>Full model predictive control can provide this foresight, but its online cost may grow quickly with horizon length, model fidelity, and the number of active geometric constraints.<n>We present a task-space receding-horizon controller that uses a short contact-consistent rollout to generate a terminal kinematic reference.
  arXiv  Detail & Related papers  (2026-07-17T08:13:12Z)
• OrthoTryOn: Geometric Orthogonalization for Conflict-Free Unified Fashion Generation [43.40052085990979]
  Unified fashion generation integrates tasks like virtual try-on and garment reconstruction into a single model to reduce task-specific adaptation costs.<n>We propose OrthoTryOn, a unified framework mitigating this interference within a shared Low- AdaptationRank (LoRA) module.
  arXiv  Detail & Related papers  (2026-06-26T09:23:25Z)
• Rethinking Multi-Condition DiTs: Eliminating Redundant Attention via Position-Alignment and Keyword-Scoping [61.459927600301654]
  Multi-condition control is bottlenecked by the conventional concatenate-and-attend'' strategy.<n>Our analysis reveals that much of this cross-modal interaction is spatially or semantically redundant.<n>We propose Position-aligned and Keyword-scoped Attention (PKA), a highly efficient framework designed to eliminate these redundancies.
  arXiv  Detail & Related papers  (2026-02-06T16:39:10Z)
• MirrorLA: Reflecting Feature Map for Vision Linear Attention [49.41670925034762]
  Linear attention significantly reduces the computational complexity of Transformers from quadratic to linear, yet it consistently lags behind softmax-based attention in performance.<n>We propose MirrorLA, a geometric framework that substitutes passive truncation with active reorientation.<n>MirrorLA achieves state-of-the-art performance across standard benchmarks, demonstrating that strictly linear efficiency can be achieved without compromising representational fidelity.
  arXiv  Detail & Related papers  (2026-02-04T09:14:09Z)
• Why Steering Works: Toward a Unified View of Language Model Parameter Dynamics [81.80010043113445]
  Local weight fine-tuning, LoRA-based adaptation, and activation-based interventions are studied in isolation.<n>We present a unified view that frames these interventions as dynamic weight updates induced by a control signal.<n>Across methods, we observe a consistent trade-off between preference and utility: stronger control increases preference while predictably reducing utility.
  arXiv  Detail & Related papers  (2026-02-02T17:04:36Z)
• Rethinking Transferable Adversarial Attacks on Point Clouds from a Compact Subspace Perspective [55.919842734983156]
  CoSA is a transferable attack framework that operates within a shared low-dimensional semantic space.<n>CoSA consistently outperforms state-of-the-art transferable attacks.
  arXiv  Detail & Related papers  (2026-01-30T15:48:11Z)
• ELLA: Efficient Lifelong Learning for Adapters in Large Language Models [12.489255789379817]
  Large Language Models (LLMs) suffer severe catastrophic forgetting when adapted sequentially to new tasks in a continual learning setting.<n>We introduce ELLA, a training framework built on the principle of selective subspace de-correlation.<n>ELLA explicitly characterizes the structure of past updates and penalizes alignments along their high-energy, task-specific directions.<n>It achieves state-of-the-art CL performance on three popular benchmarks, with relative accuracy gains of up to $9.6%$ and a $35times$ smaller memory footprint.
  arXiv  Detail & Related papers  (2026-01-05T15:58:08Z)
• ReMA: A Training-Free Plug-and-Play Mixing Augmentation for Video Behavior Recognition [11.125637599538988]
  We propose Representation-aware Mixing (ReMA), a plug-and-play augmentation strategy that formulates mixing as a controlled replacement process.<n>ReMA performs intra-class mixing under distributional constraints, suppressing irrelevant intra-class drift while enhancing statistical reliability.<n>By jointly controlling how and where mixing is applied, ReMA improves representation without additional supervision or trainable parameters.
  arXiv  Detail & Related papers  (2026-01-01T11:20:19Z)
• Drift No More? Context Equilibria in Multi-Turn LLM Interactions [58.69551510148673]
  contexts drift is the gradual divergence of a model's outputs from goal-consistent behavior across turns.<n>Unlike single-turn errors, drift unfolds temporally and is poorly captured by static evaluation metrics.<n>We show that multi-turn drift can be understood as a controllable equilibrium phenomenon rather than as inevitable decay.
  arXiv  Detail & Related papers  (2025-10-09T04:48:49Z)
• Integrating DeepRL with Robust Low-Level Control in Robotic Manipulators for Non-Repetitive Reaching Tasks [0.24578723416255746]
  In robotics, contemporary strategies are learning-based, characterized by a complex black-box nature and a lack of interpretability. We propose integrating a collision-free trajectory planner based on deep reinforcement learning (DRL) with a novel auto-tuning low-level control strategy.
  arXiv  Detail & Related papers  (2024-02-04T15:54:03Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.