Exemplar-free class incremental learning (EF-CIL) is a nontrivial task that requires continuously enriching model capability with new classes while maintaining previously learned knowledge without storing and replaying any old class exemplars. An emerging theory-guided framework for CIL trains task-specific models for a shared network, shifting the pressure of forgetting to task-id prediction. In EF-CIL, task-id prediction is more challenging due to the lack of inter-task interaction (e.g., replays of exemplars). To address this issue, we conduct a theoretical analysis of the importance and feasibility of preserving a discriminative and consistent feature space, upon which we propose a novel method termed DCNet. Concretely, it progressively maps class representations into a hyperspherical space, in which different classes are orthogonally distributed to achieve ample inter-class separation. Meanwhile, it also introduces compensatory training to adaptively adjust supervision intensity, thereby aligning the degree of intra-class aggregation. Extensive experiments and theoretical analysis verified the superiority of DCNet. Code is available at https://github.com/Tianqi-Wang1/DCNet.

Recent advances in video action recognition have achieved remarkable performance in coarse-grained macro-action classification by leveraging large-scale visual backbones and transformer architectures. However, extending these successes to fine-grained micro-action recognition remains a fundamental challenge due to the subtlety, brevity, and low motion intensity of micro-actions. In this paper, we propose a high-capacity framework for micro-action recognition, enhancing both representation learning and decision robustness. We scale to large-scale backbones using the VideoMAEv2 Giant model, enabling the extraction of finer spatial-temporal features. A Temporal-Spatial Connector (TSC) is introduced to dynamically highlight discriminative temporal frames and spatial regions, strengthening the model's focus on subtle motion cues critical for micro-action identification. To stabilize optimization and fully exploit the capacity of large models, we design a four-phase progressive training strategy, encompassing linear probing, full fine-tuning, connector-specific optimization, and classifier head refinement. Furthermore, we propose a novel ensemble decision mechanism that integrates Top-K predictions from diverse models via a Large Language Model (LLM), enhancing prediction consistency and robustness through multimodel consensus. Our method achieves an F1mean of 76.54% on the MA-52 dataset, ranking 3rd in the 2025 Micro-Action Analysis Grand Challenge and advancing the state of the art in fine-grained video understanding.

In the scenario of class-incremental learning (CIL), deep neural networks have to adapt their model parameters to non-stationary data distributions, e.g., the emergence of new classes over time. To mitigate the catastrophic forgetting phenomenon, typical CIL methods either cumulatively store exemplars of old classes for retraining model parameters from scratch or progressively expand model size as new classes arrive, which, however, compromises their practical value due to little attention paid toparameter efficiency. In this paper, we contribute a novel solution, effective control of the parameters of a well-trained model, by the synergy between two complementary learning subnetworks. Specifically, we integrate one plastic feature extractor and one analytical feed-forward classifier into a unified framework amenable to streaming data. In each CIL session, it achieves non-overwritten parameter updates in a cost-effective manner, neither revisiting old task data nor extending previously learned networks; Instead, it accommodates new tasks by attaching a tiny set of declarative parameters to its backbone, in which only one matrix per task or one vector per class is kept for knowledge retention. Experimental results on a variety of task sequences demonstrate that our method achieves competitive results against state-of-the-art CIL approaches, especially in accuracy gain, knowledge transfer, training efficiency, and task-order robustness. Furthermore, a graceful forgetting implementation on previously learned trivial tasks is empirically investigated to make its non-growing backbone (i.e., a model with limited network capacity) suffice to train on more incoming tasks.

Deep neural networks are susceptible to catastrophic forgetting when trained on sequential tasks. Various continual learning (CL) methods often rely on exemplar buffers or/and network expansion for balancing model stability and plasticity, which, however, compromises their practical value due to privacy and memory concerns. Instead, this paper considers a strict yet realistic setting, where the training data from previous tasks is unavailable and the model size remains relatively constant during sequential training. To achieve such desiderata, we propose a conceptually simple yet effective method that attributes forgetting to layer-wise parameter overwriting and the resulting decision boundary distortion. This is achieved by the synergy between two key components: HSIC-Bottleneck Orthogonalization (HBO) implements non-overwritten parameter updates mediated by Hilbert-Schmidt independence criterion in an orthogonal space and EquiAngular Embedding (EAE) enhances decision boundary adaptation between old and new tasks with predefined basis vectors. Extensive experiments demonstrate that our method achieves competitive accuracy performance, even with absolute superiority of zero exemplar buffer and 1.02x the base model.

Class-incremental learning (CIL) aims to train a model to learn new classes from non-stationary data streams without forgetting old ones. In this paper, we propose a new kind of connectionist model by tailoring neural unit dynamics that adapt the behavior of neural networks for CIL. In each training session, it introduces a supervisory mechanism to guide network expansion whose growth size is compactly commensurate with the intrinsic complexity of a newly arriving task. This constructs a near-minimal network while allowing the model to expand its capacity when cannot sufficiently hold new classes. At inference time, it automatically reactivates the required neural units to retrieve knowledge and leaves the remaining inactivated to prevent interference. We name our model AutoActivator, which is effective and scalable. To gain insights into the neural unit dynamics, we theoretically analyze the model’s convergence property via a universal approximation theorem on learning sequential mappings, which is under-explored in the CIL community. Experiments show that our method achieves strong CIL performance in rehearsal-free and minimal-expansion settings with different backbones.

Artificial neural networks are prone to suffer from catastrophic forgetting. Networks trained on something new tend to rapidly forget what was learned previously, a common phenomenon within connectionist models. In this work, we propose an effective and efficient continual learning framework using random theory, together with Bayes' rule, to equip a single model with the ability to learn streaming data. The core idea of our framework is to preserve the performance of old tasks by guiding output weights to stay in a region of low error while encountering new tasks. In contrast to the existing continual learning approaches, our main contributions concern (1) closed-formed solutions with detailed theoretical analysis; (2) training continual learners by one-pass observation of samples; (3) remarkable advantages in terms of easy implementation, efficient parameters, fast convergence, and strong task-order robustness. Comprehensive experiments under popular image classification benchmarks, FashionMNIST, CIFAR-100, and ImageNet, demonstrate that our methods predominately outperform the extensive state-of-the-art methods on training speed while maintaining superior accuracy and the number of parameters, in the class incremental learning scenario. Code is available at https://github.com/toil2sweet/CRNet.

As a class of randomized learner model, stochastic configuration networks (SCNs) have been successfully applied in a few data analytics tasks. Given the industrial big data modeling tasks, however, the original SCNs potentially lead to excessive training time. To this end, this article extends SCNs to a hybrid parallel version, termed hybrid parallel SCNs (HPSCNs). In the hybrid parallel learning algorithm, two SCNs are synchronously constructed. The difference between the two of them is that one employs a point-incremental algorithm, and another one adopts a block-incremental algorithm. Moreover, a data parallel method is established based on a dynamic block strategy to accelerate the establishment of candidate node pool for each SCN. Additionally, this article proposes an adaptive hyperparameter adjustment method, which allows the hyperparameters in the supervisory mechanism to be automatically adjusted along with the learning process. Comparative experiments are first conducted through four large-scale benchmark datasets, followed by the fully discussion on the algorithm parameters. Finally, a practical industrial application case is made, where a grinding particle size soft sensor is developed based on HPSCN, showing the effectiveness of the proposed algorithm.