Active magnetic compensation (AMC) systems are essential in precision industries, such as battery diagnostics and nondestructive testing. However, the inherent hysteretic nonlinearity of ferromagnetic shielding materials limits compensation accuracy: model-free methods struggle to capture path-dependent magnetization memory, while conventional optimal control faces fundamental challenges because hysteresis violates the Markovian assumption. This article proposes a physics-informed observer-based adaptive dynamic programming (ADP) framework with three key contributions. First, the non-Markovian nature of hysteretic AMC dynamics, which prevents direct application of Bellman’s optimality principle, is resolved by incorporating irreversible magnetization from the Jiles–Atherton (J–A) model as an augmented state variable, extending ADP applicability to memory-dependent systems. Second, the unmeasurability of this internal magnetization state is addressed through an augmented Unscented Kalman Filter that embeds J–A hysteresis dynamics into its prediction step, providing the ADP controller with complete state information. Third, closed-loop stability of the estimation-control architecture is established via Lyapunov-based input-to-state stability analysis under bounded estimation uncertainty. Experimental validation demonstrates 66.7% reduction in residual field mean absolute error compared to proportional–integral–derivative, 72.6% improvement over pseudopartial-derivative algorithm, and 24.7% enhancement over conventional ADP under multifrequency disturbances representative of industrial operating conditions.

Action recognition is a fundamental task in video understanding. Existing methods typically extract unified features to process all actions in one video, which makes it challenging to model the interactions between different objects in multi-action scenarios. To alleviate this issue, we explore disentangling any specified actions from complex scenes as an effective solution. In this paper, we propose Prompt-guided Disentangled Representation for Action Recognition (ProDA), a novel framework that disentangles any specified actions from a multi-action scene. ProDA leverages Spatio-temporal Scene Graphs (SSGs) and introduces Dynamic Prompt Module (DPM) to guide a Graph Parsing Neural Network (GPNN) in generating action-specific representations. Furthermore, we design a video-adapted GPNN that aggregates information using dynamic weights. Extensive experiments on two complex video action datasets, Charades and SportsHHI, demonstrate the effectiveness of our approach against state-of-the-art methods. Our code can be found in https://github.com/iamsnaping/ProDA.git.

Abstract Few-shot learning (FSL) aims at distilling transferable knowledge on existing concepts to cope with novel concepts for which only a few labeled data are available. Most of the popular FSL methods acquire this knowledge by learning on large-scale supervised data from the existing concepts. Considering obtaining supervised data might sometimes be difficult and heavy-burden, we pursue a relatively mild prerequisite for FSL, that is, using unsupervised instead of supervised data to acquire the transferable knowledge. We propose a novel easy-to-implement FSL framework, U nsupervised T ractive M omentum (UTM), composed of modular dual encoders, a combinatorial loss mechanism, and a classifier that together form a reusable and extensible learning system, that only requires unsupervised data of existing concepts. UTM randomly samples unsupervised data and augments them to create many synthetic query-key matching tasks on-the-fly, and deploys two different encoders while possessing identical architecture, named traction encoder and momentum encoder , to learn a representation space by a combinatorial parameter updating manner. The representation space learned on unsupervised data is expected to be a good fit to few-shot recognition on novel concepts. UTM is composed of parallelizable dual encoders and optimized for scalable training in GPU-based high-performance computing environments. Theoretical convergence and bound analysis further support its deployment in distributed systems. Theoretical justifications of the parameter updating mechanism in UTM are given from the perspective of convergence, and a theoretical loss bound for UTM is proved, which mathematically quantifies the relationship between our self-supervised UTM and the vanilla supervised method. Extensive experimental evaluation on several benchmark datasets demonstrates that UTM yields significant improvement to state-of-the-art unsupervised methods even very close to supervised methods, which can also be well explained using our theory.

Building learning systems possessing adaptive flexibility to different tasks is critical and challenging. In this article, we propose a novel and general meta-learning framework, called meta-modulation (MeMo), to foster the adaptation capability of a base learner across different tasks where only a few training data are available per task. For one independent task, MeMo proceeds like a "feedback regulation system," which achieves an adaptive modulation on the so-called definitive embeddings of query data to maximize the corresponding task objective. Specifically, we devise a type of efficient feedback information, definitive embedding feedback (DEF), to mathematize and quantify the unsuitability between the few training data and the base learner as well as the promising adjustment direction to reduce this unsuitability. The DEFs are encoded into high-level representation and temporarily stored as task-specific modulator templates by a modulation encoder. For coming query data, we develop an attention mechanism acting upon these modulator templates and combine both task/data-level modulation to generate the final data-specific meta-modulator. This meta-modulator is then used to modulate the query's embedding for correct decision-making. Our framework is scalable for various base learner models like multi-layer perceptron (MLP), long short-term memory (LSTM), convolutional neural network (CNN), and transformer, and applicable to different learning problems like language modeling and image recognition. Experimental results on a 2-D point synthetic dataset and various benchmarks in language and vision domains demonstrate the effectiveness and competitiveness of our framework.

Remarkable gains in deep learning usually benefit from large-scale supervised data. Ensuring the intra-class modality diversity in training set is critical for generalization capability of cutting-edge deep models, but it burdens human with heavy manual labor on data collection and annotation. In addition, some rare or unexpected modalities are new for the current model, causing reduced performance under such emerging modalities. Inspired by the achievements in speech recognition, psychology and behavioristics, we present a practical solution, self-reinforcing unsupervised matching (SUM), to annotate the images with 2D structure-preserving property in an emerging modality by cross-modality matching. Specifically, we propose a dynamic programming algorithm, dynamic position warping (DPW), to reveal the underlying element correspondence relationship between two matrix-form data in an order-preserving fashion, and devise a local feature adapter (LoFA) to allow for cross-modality similarity measurement. On these bases, we develop a two-tier self-reinforcing learning mechanism on both feature level and image level to optimize the LoFA. The proposed SUM framework requires no any supervision in emerging modality and only one template in seen modality, providing a promising route towards incremental learning and continual learning. Extensive experimental evaluation on two proposed challenging one-template visual matching tasks demonstrate its efficiency and superiority.

Few-shot learning (FSL) focuses on distilling transferrable knowledge from existing experience to cope with novel concepts for which the labeled data are scarce. A typical assumption in FSL is that the training examples of novel classes are all clean with no outlier interference. In many realistic applications where examples are provided by users, however, data are potentially noisy or unreadable. In this context, we introduce a novel research topic, robust FSL (RFSL), where we aim to address two types of outliers within user-provided data: the representation outlier (RO) and the label outlier (LO). Moreover, we introduce a metric for estimating robustness and use it to investigate the performance of several advanced methods to FSL when faced with user-provided outliers. In addition, we propose robust attentive profile networks (RapNets) to achieve outlier suppression. The results of a comprehensive evaluation of benchmark data sets demonstrate the shortcomings of current FSL methods and the superiority of the proposed RapNets when dealing with RFSL problems, establishing a benchmark for follow-up studies.

Human has the ability of drawing inferences about other things from only one instance. Few-shot learning is aimed at imitating this generalized learning behavior of human beings, where the learning machine is expected to recognize novel categories not seen in the training set, given only a few training data for each novel category. In this paper, we enhance the Prototypical Network for few-shot learning tasks by introducing a domain alignment module, which takes into account the domain shifts existing between different categories. Compared to original Prototypical Network (PN), the most excellent model for few-shot learning at present, our proposed Domain Alignment Prototypical Network (DA-PN) is able to abate the distribution differences among the data of training and test classes, further optimizing the embedding space of prototype feature for each category and then boosting few-shot recognition. Comprehensive empirical evidence demonstrates that the proposed DA-PN can yield state-of-the-art few-shot recognition performance on the public benchmark dataset mini-ImageNet as well as a novel proposed few-shot dataset MNIST&CIFAR10.

In this paper, we propose to build a comprehensive cyberphysical system (CPS) with virtual reality (VR) and intelligent sensors for motion recognition and training. We use both wearable wireless sensors (such as electrocardiogram, motion sensors) and nonintrusive wireless sensors (such as gait sensors) to monitor the motion training status. We first provide our CPS architecture. Then we focus on motion training from three perspectives: 1) VR-first we introduce how we can use motion capture camera to trace the motions; 2) gait recognition-we have invented low-cost small wireless pyroelectric sensor, which can recognize different gaits through Bayesian pattern learning. It can automatically measure gait training effects; and 3) gesture recognition-to quickly tell what motions the subject is doing, we propose a low-cost, low-complexity motion recognition system with 3-axis accelerometers. We will provide hardware and software design. Our experimental results validate the efficiency and accuracy of our CPS design.

This paper presents an information-gain-based sensor selection approach as well as a sensor sensing probability model-based calibration process for multihuman tracking in distributed binary pyroelectric infrared sensor networks. This research includes three contributions: 1) choose the subset of sensors that can maximize the mutual information between sensors and targets; 2) find the sensor sensing probability model to represent the sensing space for sensor calibration; and 3) provide a factor graph-based message passing scheme for distributed tracking. Our approach can find the solution for sensor selection to optimize the performance of tracking. The sensing probability model is efficiently optimized through the calibration process in order to update the parameters of sensor positions and rotations. An application for mobile calibration and tracking is developed. Simulation and experimental results are provided to validate the proposed framework.

Context-aware applications rely on contexts derived from sensory data to adapt their behavior. However, contexts can be inconsistent and cause application anomaly or crash. One popular solution is to detect and resolve context inconsistencies at runtime. However, we observe that many detected inconsistencies do not indicate real context problems. Instead, they are caused by improper inconsistency detection. These inconsistencies are harmless, and their resolution is unnecessary or may even cause new problems. We name them inconsistency hazards. Inconsistency hazards should be suppressed, but their occurrences resemble normal inconsistencies. In this paper, we present a pattern-learning based approach SHAP to suppressing the detection of inconsistency hazards. Our key insight is that the detection of such hazards is subject to certain patterns of context changes. These patterns, although difficult to specify manually, can be learned effectively from historical inconsistency detection data. We evaluated our SHAP experimentally through three context-aware applications. The results reported that SHAP can automatically suppress the detection of over 90% inconsistency hazards, while preserving the detection of over 98% normal inconsistencies, with only negligible overhead.

In this paper, we present a framework for complex scenarios recognition with crowded walkers. This study aims to develop an alternative surveillance system to traditional video camera and visual sensor based systems. Instead of utilizing visual devices in traditional surveillance systems, our crowded scene analysis is based on PIR (Pyroelectric Infrared) sensor networks with intelligent algorithms for context pattern extraction and analysis. Specifically, we will propose two new ideas to handle the crowded scenes: (1) Use hierarchical Bayesian NMF (Non-negative Matrix Factorization) algorithm to automatically identify the basic pattern basis, which will be used for accurate scenario recognition, (2) Use a tree-based structure to organize all basic features for fast object recognition. The experimental results valid the efficiency of the proposed two schemes on crowded scenario recognition with low-cost, non-visual system. The results also demonstrate that our framework is appropriate to be implemented in a wireless sensor based monitoring system under severe circumstances.