Explore the vast range of titles available.

Acoustic scene analysis (ASA) relies on the dynamic sensing and understanding of stationary and non-stationary sounds from various events, background noises and human actions with objects. However, the spatio-temporal nature of the sound signals may not be stationary, and novel events may exist that eventually deteriorate the performance of the analysis. In this study, a self-learning-based ASA for acoustic event recognition (AER) is presented to detect and incrementally learn novel acoustic events by tackling catastrophic forgetting. The proposed ASA framework comprises six elements: (1) raw acoustic signal pre-processing, (2) low-level and deep audio feature extraction, (3) acoustic novelty detection (AND), (4) acoustic signal augmentations, (5) incremental class-learning (ICL) (of the audio features of the novel events) and (6) AER. The self-learning on different types of audio features extracted from the acoustic signals of various events occurs without human supervision. For the extraction of deep audio representations, in addition to visual geometry group (VGG) and residual neural network (ResNet), time-delay neural network (TDNN) and TDNN based long short-term memory (TDNN–LSTM) networks are pre-trained using a large-scale audio dataset, Google AudioSet. The performances of ICL with AND using Mel-spectrograms, and deep features with TDNNs, VGG, and ResNet from the Mel-spectrograms are validated on benchmark audio datasets such as ESC-10, ESC-50, UrbanSound8K (US8K), and an audio dataset collected by the authors in a real domestic environment.

Deep neural networks (DNNs) have drawn much attention due to their success in various vision tasks. Current DNNs are used on data with a relatively small number of classes (e.g., 1000 or less) and employ a fully connected layer for classification, which allocates one neuron for each class and thus, per-example, the classification scales as O(K) with the number of classes K. This approach is computationally intensive for many real-life applications where the number of classes is very large (e.g., tens of thousands of classes). To address this problem, our paper introduces a hierarchical approach for classification with a large number of classes that scales as O(K) and could be extended to O(logK) with a deeper hierarchy. The method, called Hierarchical PPCA, uses a self-supervised pretrained feature extractor to obtain meaningful features and trains Probabilistic PCA models on the extracted features for each class separately, making it easy to add classes without retraining the whole model. The Mahalanobis distance is used to obtain the classification result. To speed-up classification, the proposed Hierarchical PPCA framework clusters the image class models, represented as Gaussians, into a smaller number of super-classes using a modified k-means clustering algorithm. The classification speed increase is obtained by Hierarchical PPCA assigning a sample to a small number of the most likely super-classes and restricting the image classification to the image classes corresponding to these super-classes. The fact that the model is trained on each class separately makes it applicable to training on very large datasets such as the whole ImageNet with more than 10,000 classes. Experiments on three standard datasets (ImageNet-100, ImageNet-1k,and ImageNet-10k) indicate that the hierarchical classifier can achieve a superior accuracy with up to a 16-fold speed increase compared to a standard fully connected classifier.

In our ever-evolving world, new data exhibits a long-tailed distribution, such as emerging images in varying amounts. This necessitates continuous model learning imbalanced data without forgetting, addressing the challenge of long-tailed class-incremental learning (LTCIL). Existing methods often rely on retraining linear classifiers with former data, which is impractical in real-world settings. In this paper, we harness the potent representation capabilities of pre-trained models and introduce AdaPtive Adapter RouTing ( Apart ) as an exemplar-free solution for LTCIL. To counteract forgetting, we train inserted adapters with frozen pre-trained weights for deeper adaptation and maintain a pool of adapters for selection during sequential model updates. Additionally, we present an auxiliary adapter pool designed for effective generalization, especially on minority classes. Adaptive instance routing across these pools captures crucial correlations, facilitating a comprehensive representation of all classes. Consequently, Apart tackles the imbalance problem as well as catastrophic forgetting in a unified framework. Extensive benchmark experiments validate the effectiveness of Apart .

Class-incremental learning (CIL) aims to adapt to emerging new classes without forgetting old ones. Traditional CIL models are trained from scratch to continually acquire knowledge as data evolves. Recently, pre-training has achieved substantial progress, making vast pre-trained models (PTMs) accessible for CIL. Contrary to traditional methods, PTMs possess generalizable embeddings, which can be easily transferred for CIL. In this work, we revisit CIL with PTMs and argue that the core factors in CIL are adaptivity for model updating and generalizability for knowledge transferring. (1) We first reveal that frozen PTM can already provide generalizable embeddings for CIL. Surprisingly, a simple baseline (SimpleCIL) which continually sets the classifiers of PTM to prototype features can beat state-of-the-art even without training on the downstream task. (2) Due to the distribution gap between pre-trained and downstream datasets, PTM can be further cultivated with adaptivity via model adaptation. We propose AdaPt and mERge ( Aper ), which aggregates the embeddings of PTM and adapted models for classifier construction. Aper is a general framework that can be orthogonally combined with any parameter-efficient tuning method, which holds the advantages of PTM’s generalizability and adapted model’s adaptivity. (3) Additionally, considering previous ImageNet-based benchmarks are unsuitable in the era of PTM due to data overlapping, we propose four new benchmarks for assessment, namely ImageNet-A, ObjectNet, OmniBenchmark, and VTAB. Extensive experiments validate the effectiveness of Aper with a unified and concise framework. Code is available at https://github.com/zhoudw-zdw/RevisitingCIL .

Incremental class learning aims to continuously learn new classes without accessing data from previous tasks. A key challenge in this setting is catastrophic forgetting, where the model adapts to new data but loses performance on old tasks. Most prior work addresses this by regularization or replay mechanisms. In this paper, we observe a consistent and underexplored issue: the classifier trained on new tasks often exhibits overconfidence on old-task samples, assigning them low predictive uncertainty. This overconfidence leads to misclassifying old samples as new classes. To mitigate this, we propose an uncertainty-calibrated loss that explicitly penalizes overconfident predictions from the new-task classifier. By integrating a calibration-aware regularization term, our method effectively encourages better uncertainty estimation without modifying the core learning framework. Importantly, our approach is plug-and-play and can be applied to a wide range of existing incremental learning methods. Extensive experiments on Sequential CIFAR-100, Sequential Tiny-ImageNet, and Sequential mini-ImageNet demonstrate that our method consistently improves existing baselines by 1%–3% in average accuracy.

Incremental learning is an emerging machine-learning approach designed to prevent catastrophic forgetting while learning a new task with knowledge retained from previous tasks. However, existing methods often must fully utilize the distributional information of the old task model’s outputs. To alleviate this, we propose a class incremental learning method grounded in evidence theory to leverage the distributional information of outputs from the old task model, which integrates uncertainty estimation into the knowledge distillation process to ensure the new task model effectively learns from the old task model’s distribution information. Specifically, we incorporate uncertainty estimation based on evidence theory to calculate knowledge distillation loss during training. We compute the uncertainty estimates of outputs from both the new and old task models, and then they are used in the knowledge distillation loss calculation. We propose a novel classification strategy that considers outputs’ probability and uncertainty estimates, which could determine the sample category without requiring an additional training phase or supplementary models. Experiments on the CIFAR-100 and ImageNet datasets demonstrate the effectiveness of the proposed method, showing that it outperforms existing methods by improving accuracy by 1%-2%. Results suggest that leveraging the distribution information of model outputs can effectively mitigate catastrophic forgetting in deep learning models within incremental learning scenarios.

In practical applications of deep learning-based Synthetic Aperture Radar (SAR) Automatic Target Recognition (ATR) systems, new target categories emerge continuously. This requires the systems to learn incrementally—acquiring new knowledge while retaining previously learned information. To mitigate catastrophic forgetting in Class-Incremental Learning (CIL), this paper proposes a CIL method for SAR ATR named Multi-center Prototype Feature Distribution Reconstruction (MPFR). It has two core components. First, a Multi-scale Hybrid Attention feature extractor is designed. Trained via a feature space optimization strategy, it fuses and extracts discriminative features from both SAR amplitude images and Attribute Scattering Center data, while preserving feature space capacity for new classes. Second, each class is represented by multiple prototypes to capture complex feature distributions. Old class knowledge is retained by modeling their feature distributions through parameterized Gaussian diffusion, alleviating feature confusion in incremental phases. Experiments on public SAR datasets show MPFR achieves superior performance compared to existing approaches, including recent SAR-specific CIL methods. Ablation studies validate each component’s contribution, confirming MPFR’s effectiveness in addressing CIL for SAR ATR without storing historical raw data.

Deep learning (DL)-based multi-user physical layer authentication (PLA) in the Industrial Internet of Things (IIoT) requires frequent updates as new users join. Class incremental learning (CIL) addresses this challenge, but existing generative replay approaches depend on heavy parameterized models, causing high computational overhead and limiting deployment in resource-constrained environments. To address these challenges, we propose a parameter-free statistical generator-based CIL framework, PSG-CIL, for DL-based multi-user PLA in the IIoT. The parameter-free statistical generator (PSG) produces Gaussian sampling on user-specific means and variances to generate pseudo-data without training extra models, greatly reducing computational overhead. A confidence-based pseudo-data selection ensures pseudo-data reliability, while a dynamic adjustment mechanism for the loss weight balances the retention of old users’ knowledge and the adaptation to new users. Experiments on real industrial datasets show that PSG-CIL consistently achieves superior accuracy while maintaining a lightweight scale; for example, in the AAP outer loop scenario, PSG-CIL reaches 70.68%, outperforming retraining from scratch (58.57%) and other CIL methods.

The human brain is capable of learning tasks sequentially mostly without forgetting. However, deep neural networks (DNNs) suffer from catastrophic forgetting when learning one task after another. We address this challenge considering a class-incremental learning scenario where the DNN sees test data without knowing the task from which this data originates. During training, Continual Prune-and-Select (CP&S) finds a subnetwork within the DNN that is responsible for solving a given task. Then, during inference, CP&S selects the correct subnetwork to make predictions for that task. A new task is learned by training available neuronal connections of the DNN (previously untrained) to create a new subnetwork by pruning, which can include previously trained connections belonging to other subnetwork(s) because it does not update shared connections. This enables to eliminate catastrophic forgetting by creating specialized regions in the DNN that do not conflict with each other while still allowing knowledge transfer across them. The CP&S strategy is implemented with different subnetwork selection strategies, revealing superior performance to state-of-the-art continual learning methods tested on various datasets (CIFAR-100, CUB-200-2011, ImageNet-100 and ImageNet-1000). In particular, CP&S is capable of sequentially learning 10 tasks from ImageNet-1000 keeping an accuracy around 94% with negligible forgetting, a first-of-its-kind result in class-incremental learning. To the best of the authors’ knowledge, this represents an improvement in accuracy above 10% when compared to the best alternative method.

The identification of surface damage and structural components is critical for structural health monitoring (SHM) in order to evaluate building safety. Recently, deep neural networks (DNNs)–based approaches have emerged rapidly. However, the existing approaches often encounter catastrophic forgetting when the trained model is used to learn new classes of interest. Conventionally, joint training of the network on both the previous and new data is employed, which is time‐consuming and demanding for computation and memory storage. To address this issue, we propose a new approach that integrates two continual learning (CL) algorithms, i.e., elastic weight consolidation (EWC) and learning without forgetting (LwF), denoted as EWCLwF. We also investigate two scenarios for a comprehensive discussion, incrementally learning the classes with similar versus dissimilar data characteristics. Results have demonstrated that EWCLwF requires significantly less training time and data storage compared to joint training, and the average accuracy is enhanced by 0.7%–4.5% compared against other baseline references in both scenarios. Furthermore, our findings reveal that all CL‐based approaches benefit from similar data characteristics, while joint training not only fails to benefit but performs even worse, which indicates a scenario that can emphasize the advantage of our proposed approach. The outcome of this study will enhance the long‐term monitoring of progressively increasing learning classes in SHM, leading to more efficient usage and management of computing resources.