Dynamic Distillation and Gradient Consistency for Robust Long-Tailed Incremental Learning

• URL: http://arxiv.org/abs/2605.03364v1
• Date: Tue, 05 May 2026 04:50:06 GMT
• Title: Dynamic Distillation and Gradient Consistency for Robust Long-Tailed Incremental Learning
• Abstract summary: We introduce consistency gradient regularization to suppress abrupt fluctuations and stabilize the training process.<n>Second, we dynamically adjust the weight of the distillation loss by measuring the degree of class imbalance with normalized entropy.<n>Experiments on the CIFAR-100-LT, ImageNetSubset-LT, and Food101-LT benchmarks show that our method achieves consistent accuracy improvements of up to 5.0%.
• Score: 19.30816824251175
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: The task of Long-tailed Class Incremental Learning (LT-CIL) addresses the sequential learning of new classes from datasets with imbalanced class distributions. This scenario intensifies the fundamental problem of catastrophic forgetting, inherent to continual learning, with the dual challenges of under-learning minority classes and overfitting majority classes. To tackle these combined issues, this paper proposes two main techniques. First, we introduce gradient consistency regularization, which leverages the moving average of gradients to suppress abrupt fluctuations and stabilize the training process. Second, we dynamically adjust the weight of the distillation loss by measuring the degree of class imbalance with normalized entropy. This adaptive weighting establishes an optimal balance between retaining old knowledge and acquiring new information. Experiments on the CIFAR-100-LT, ImageNetSubset-LT, and Food101-LT benchmarks show that our method achieves consistent accuracy improvements of up to 5.0\%. Furthermore, we demonstrate dramatic gains in the challenging 'In-ordered' setting, where tasks progress from majority to minority classes, highlighting our method's robustness in mitigating forgetting under unfavorable learning dynamics. This enhanced performance is achieved without a significant increase in computational overhead, demonstrating the practicality of our framework.

Related papers

• Stabilizing Reinforcement Learning with LLMs: Formulation and Practices [61.361819972410046]
  We show why and under what conditions the true sequence-level reward can be optimized via a surrogate token-level objective in policy gradient methods such as REINFORCE.<n>This insight provides a principled explanation for the crucial role of several widely adopted techniques in stabilizing RL training.
  arXiv  Detail & Related papers  (2025-12-01T07:45:39Z)
• AdaSwitch: Adaptive Switching Generation for Knowledge Distillation [58.647880811071495]
  Small language models (SLMs) are crucial for applications with strict latency and computational constraints.<n>We propose AdaSwitch, a novel approach that combines on-policy and off-policy generation at the token level.<n>AdaSwitch consistently improves accuracy, offering a practical and effective method for distilling SLMs with acceptable additional overhead.
  arXiv  Detail & Related papers  (2025-10-09T06:38:37Z)
• Sculpting Subspaces: Constrained Full Fine-Tuning in LLMs for Continual Learning [19.27175827358111]
  Continual learning in large language models (LLMs) is prone to catastrophic forgetting, where adapting to new tasks significantly degrades performance on previously learned ones.<n>We propose a novel continual full fine-tuning approach leveraging adaptive singular value decomposition (SVD)<n>We evaluate our approach extensively on standard continual learning benchmarks using both encoder-decoder (T5-Large) and decoder-only (LLaMA-2 7B) models.
  arXiv  Detail & Related papers  (2025-04-09T17:59:42Z)
• SLCA++: Unleash the Power of Sequential Fine-tuning for Continual Learning with Pre-training [68.7896349660824]
  We present an in-depth analysis of the progressive overfitting problem from the lens of Seq FT. Considering that the overly fast representation learning and the biased classification layer constitute this particular problem, we introduce the advanced Slow Learner with Alignment (S++) framework. Our approach involves a Slow Learner to selectively reduce the learning rate of backbone parameters, and a Alignment to align the disjoint classification layers in a post-hoc fashion.
  arXiv  Detail & Related papers  (2024-08-15T17:50:07Z)
• Learning Continually by Spectral Regularization [45.55508032009977]
  Continual learning algorithms seek to mitigate loss of plasticity by sustaining good performance while maintaining network trainability. We develop a new technique for improving continual learning inspired by the observation that the singular values of the neural network parameters at initialization are an important factor for trainability during early phases of learning. We present an experimental analysis that shows how the proposed spectral regularizer can sustain trainability and performance across a range of model architectures in continual supervised and reinforcement learning settings.
  arXiv  Detail & Related papers  (2024-06-10T21:34:43Z)
• Gradient Reweighting: Towards Imbalanced Class-Incremental Learning [8.438092346233054]
  Class-Incremental Learning (CIL) trains a model to continually recognize new classes from non-stationary data. A major challenge of CIL arises when applying to real-world data characterized by non-uniform distribution. We show that this dual imbalance issue causes skewed gradient updates with biased weights in FC layers, thus inducing over/under-fitting and catastrophic forgetting in CIL.
  arXiv  Detail & Related papers  (2024-02-28T18:08:03Z)
• Bias Mitigating Few-Shot Class-Incremental Learning [17.185744533050116]
  Few-shot class-incremental learning aims at recognizing novel classes continually with limited novel class samples. Recent methods somewhat alleviate the accuracy imbalance between base and incremental classes by fine-tuning the feature extractor in the incremental sessions. We propose a novel method to mitigate model bias of the FSCIL problem during training and inference processes.
  arXiv  Detail & Related papers  (2024-02-01T10:37:41Z)
• Class Gradient Projection For Continual Learning [99.105266615448]
  Catastrophic forgetting is one of the most critical challenges in Continual Learning (CL) We propose Class Gradient Projection (CGP), which calculates the gradient subspace from individual classes rather than tasks.
  arXiv  Detail & Related papers  (2023-11-25T02:45:56Z)
• Beyond Losses Reweighting: Empowering Multi-Task Learning via the Generalization Perspective [61.10883077161432]
  Multi-task learning (MTL) trains deep neural networks to optimize several objectives simultaneously using a shared backbone.<n>We introduce a novel MTL framework that leverages weight perturbation to regulate gradient norms, thus improving generalization.<n>Our method significantly outperforms existing gradient-based MTL techniques in terms of task performance and overall model robustness.
  arXiv  Detail & Related papers  (2022-11-24T17:19:30Z)
• A Theoretical Analysis of the Learning Dynamics under Class Imbalance [0.10231119246773925]
  We show that the learning curves for minority and majority classes follow sub-optimal trajectories when training with a gradient-based trajectory. This slowdown is related to the imbalance ratio and can be traced back to a competition between the optimization of different classes. We find that GD is not guaranteed to decrease the loss for each class but that this problem can be addressed by performing a per-class normalization of the gradient.
  arXiv  Detail & Related papers  (2022-07-01T12:54:38Z)
• Class-incremental Learning with Rectified Feature-Graph Preservation [24.098892115785066]
  A central theme of this paper is to learn new classes that arrive in sequential phases over time. We propose a weighted-Euclidean regularization for old knowledge preservation. We show how it can work with binary cross-entropy to increase class separation for effective learning of new classes.
  arXiv  Detail & Related papers  (2020-12-15T07:26:04Z)
• Extrapolation for Large-batch Training in Deep Learning [72.61259487233214]
  We show that a host of variations can be covered in a unified framework that we propose. We prove the convergence of this novel scheme and rigorously evaluate its empirical performance on ResNet, LSTM, and Transformer.
  arXiv  Detail & Related papers  (2020-06-10T08:22:41Z)

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.