Distribution Mismatch Research Articles

Joint Energy-based Model for Semi-supervised Respiratory Sound Classification: A Method of Insensitive to Distribution Mismatch.

Semi-supervised learning effectively mitigates the lack of labeled data by introducing extensive unlabeled data. Despite achieving success in respiratory sound classification, in practice, it usually takes years to acquire a sufficiently sizeable unlabeled set, which consequently results in an extension of the research timeline. Considering that there are also respiratory sounds available in other related tasks, like breath phase detection and COVID-19 detection, it might be an alternative manner to treat these external samples as unlabeled data for respiratory sound classification. However, since these external samples are collected in different scenarios via different devices, there inevitably exists a distribution mismatch between the labeled and external unlabeled data. For existing methods, they usually assume that the labeled and unlabeled data follow the same data distribution. Therefore, they cannot benefit from external samples. To utilize external unlabeled data, we propose a semi-supervised method based on Joint Energy-based Model (JEM) in this paper. During training, the method attempts to use only the essential semantic components within the samples to model the data distribution. When non-semantic components like recording environments and devices vary, as these non-semantic components have a small impact on the model training, a relatively accurate distribution estimation is obtained. Therefore, the method exhibits insensitivity to the distribution mismatch, enabling the model to leverage external unlabeled data to mitigate the lack of labeled data. Taking ICBHI 2017 as the labeled set, HF_Lung_V1 and COVID-19 Sounds as the external unlabeled sets, the proposed method exceeds the baseline by 12.86.

Across Sessions and Subjects Domain Adaptation for Building Robust Myoelectric Interface.

Gesture interaction via surface electromyography (sEMG) signal is a promising approach for advanced human-computer interaction systems. However, improving the performance of the myoelectric interface is challenging due to the domain shift caused by the signal's inherent variability. To enhance the interface's robustness, we propose a novel adaptive information fusion neural network (AIFNN) framework, which could effectively reduce the effects of multiple scenarios. Specifically, domain adversarial training is established to inhibit the shared network's weights from exploiting domain-specific representation, thus allowing for the extraction of domain-invariant features. Effectively, classification loss, domain diversence loss and domain discrimination loss are employed, which improve classification performance while reduce distribution mismatches between the two domains. To simulate the application of myoelectric interface, experiments were carried out involving three scenarios (intra-session, inter-session and inter-subject scenarios). Ten non-disabled subjects were recruited to perform sixteen gestures for ten consecutive days. The experimental results indicated that the performance of AIFNN was better than two other state-of-the-art transfer learning approaches, namely fine-tuning (FT) and domain adversarial network (DANN). This study demonstrates the capability of AIFNN to maintain robustness over time and generalize across users in practical myoelectric interface implementations. These findings could serve as a foundation for future deployments.

Improving Human Activity Recognition With Wearable Sensors Through BEE: Leveraging Early Exit and Gradient Boosting.

Early-exiting has recently provided an ideal solution for accelerating activity inference by attaching internal classifiers to deep neural networks. It allows easy activity samples to be predicted at shallower layers, without executing deeper layers, hence leading to notable adaptiveness in terms of accuracy-speed trade-off under varying resource demands. However, prior most works typically optimize all the classifiers equally on all types of activity data. As a result, deeper classifiers will only see hard samples during test phase, which renders the model suboptimal due to the training-test data distribution mismatch. Such issue has been rarely explored in the context of activity recognition. In this paper, to close the gap, we propose to organize all these classifiers as a dynamic-depth network and jointly optimize them in a similar gradient-boosting manner. Specifically, a gradient-rescaling is employed to bound the gradients of parameters at different depths, that makes such training procedure more stable. Particularly, we perform a prediction reweighting to emphasize current deep classifier while weakening the ensemble of its previous classifiers, so as to relieve the shortage of training data at deeper classifiers. Comprehensive experiments on multiple HAR benchmarks including UCI-HAR, PAMAP2, UniMiB-SHAR, and USC-HAD verify that it is state-of-the-art in accuracy and speed. A real implementation is measured on an ARM-based mobile device.

A Data-Driven Framework for Power System Event Type Identification via Safe Semi-Supervised Techniques

This paper investigates the use of phasor measurement unit (PMU) data with deep learning techniques to construct real-time event identification models for transmission networks. Increasing penetration of distributed energy resources represents a great opportunity to achieve decarbonization, as well as challenges in systematic situational awareness. When high-resolution PMU data and sufficient manually recorded event labels are available, the power event identification problem is defined as a statistical classification problem that can be solved by numerous cutting-edge classifiers. However, in real grids, collecting tremendous high-quality event labels is quite expensive. Utilities frequently have a large number of event records without in-depth details (i.e., unlabeled events). To bridge this gap, we propose a novel semi-supervised learning-based method to improve the performance of event classifiers trained with a limited number of labeled events by exploiting the information from massive unlabeled events. In other words, compared to existing data-driven methods, our method requires only a small portion of labeled data to achieve a similar level of accuracy. Meanwhile, this work discusses and addresses the performance degradation caused by class distribution mismatch between the training set and the real applications. Specifically, this method utilizes pseudo-labeling technique to investigate the value of unlabeled events and incrementally expands the training dataset. Moreover, a safe learning mechanism is developed to mitigate the impacts of class distribution mismatch and prevent performance degradation. Based on the proposed safe learning mechanism, our model does not directly use all unlabeled events during model training, but selectively uses them through a comprehensive evaluation procedure. Numerical studies on a sizable PMU dataset have been used to validate the performance of the proposed method.

Self-supervised learning minimax entropy domain adaptation for the underwater target recognition

With wide research of intelligent methods, studies on underwater target recognition have been making rapid progress. However, various marine conditions may cause data distribution mismatch between the collected signal sets, reducing model recognition performance. To mitigate the negative impact of data divergence, this paper uses the domain adaptation methods in target recognition and proposes an improved domain adaptation frame, self-supervised learning minimax entropy. Firstly, based on the minimax entropy method (MME), the prediction consistency is utilized to determine pseudo-labels, and the loss weight is introduced to deal with the misaligned target domain data. Then, a self-supervised learning mechanism is designed to ensure consistency of prediction results during training. Three different features, including the constant-Q transform (CQT), Mel spectrum, and Mel-frequency cepstral coefficient (MFCC), are used to verify the performance of domain adaptation methods. The experimental results show that applying domain adaptations can effectively improve the recognition performance of the models under most experimental conditions, and the improved frame has higher average recognition accuracy than other domain adaptation methods in the experiments.

Enhanced 3D Pose Estimation in Multi-Person, Multi-View Scenarios through Unsupervised Domain Adaptation with Dropout Discriminator.

Data-driven pose estimation methods often assume equal distributions between training and test data. However, in reality, this assumption does not always hold true, leading to significant performance degradation due to distribution mismatches. In this study, our objective is to enhance the cross-domain robustness of multi-view, multi-person 3D pose estimation. We tackle the domain shift challenge through three key approaches: (1) A domain adaptation component is introduced to improve estimation accuracy for specific target domains. (2) By incorporating a dropout mechanism, we train a more reliable model tailored to the target domain. (3) Transferable Parameter Learning is employed to retain crucial parameters for learning domain-invariant data. The foundation for these approaches lies in the H-divergence theory and the lottery ticket hypothesis, which are realized through adversarial training by learning domain classifiers. Our proposed methodology is evaluated using three datasets: Panoptic, Shelf, and Campus, allowing us to assess its efficacy in addressing domain shifts in multi-view, multi-person pose estimation. Both qualitative and quantitative experiments demonstrate that our algorithm performs well in two different domain shift scenarios.

Estimating water pollution and economic cost embodied in the mining industry: An interprovincial analysis in China

China's rapid industrialization since the 21st century has led to significant mineral resource consumption, causing a dual dilemma of water scarcity and pollution. Imbalanced supply and demand of mineral resources from interregional trade, combined with spatial distribution mismatches of water and mineral resources, have led to the reconstruction of water pollution discharge patterns. This study employs a multiregional input-output model and opportunity cost and restoration cost methods to assess the transfer of embodied water pollution in China's mining industry and the resulting economic losses. Findings indicate significant environmental inequities across the country, with the Beijing-Tianjin area and eastern coastal regions responsible for water pollution spillover and cost passing to other regions. The central and western regions bear a greater water pollution transfer and control burden. This research on embodied water pollution transfer in mineral trade is crucial for national water environment governance planning, with practical implications for rational water resource redistribution, improved ecological compensation, and coordinated regional economic development.

Decomposing texture and semantic for out-of-distribution detection

The out-of-distribution (OOD) detection task assumes samples that follow the distribution of training data as in-distribution (ID), while samples from other data distributions are considered OOD. In recent years, the OOD detection tasks have made significant progress since many studies observed that the distribution mismatch between training and real datasets can severely deteriorate the reliability of AI systems. Nevertheless, the lack of precise interpretation for the in-distribution (ID) limits the application of the OOD detection methods to real-world systems. To tackle this, we decompose the definition of the ID into texture and semantics, motivated by the demands of real-world scenarios. We also design new benchmarks to measure the robustness that OOD detection methods should have. Our proposed benchmark verifies not only the precision but also the robustness of the detection models. It is crucial to measure both factors in OOD detection as they indicate different traits of the model. For instance, precision is relevant to scenarios that detect minor cracks in the conveyor belt of a smart factory, whereas robustness pertains to maintaining performance under diverse weather conditions, as required by autonomous driving. To achieve a good balance between the OOD detection performance and robustness, our method takes a divide-and-conquer approach. Specifically, the proposed model first handles each component of the texture and semantics separately and then fuses these later. This philosophy is empirically proven by a series of benchmarks including both the proposed and the conventional counterpart. By decomposing the prior “unclear” definition of the ID into texture and semantic components, our novel approach better suits the demands of a reliable machine learning system, which requires robustness and consistent performance across varied scenarios. Unlike prior works, our approach does not rely on any extra datasets or labels. This prevents our proposed framework from being dependent on a particular dataset distribution.

Unsupervised Domain Adaptation with CycleGAN: Adapting Image Style and Content for Improved Cross-Domain Performance

Abstract: In order to improve model performance while transferring information from a source domain to an unlabelled target domain, unsupervised domain adaptation is a crucial problem in deep learning. The CycleGAN design, a well-known generative adversarial network architecture, is used in this study to explore the domain adaptation challenge. Improved cross-domain performance is the goal, and both picture style and content adaptation are being focused on. In this study, we explore the use of CycleGAN to modify images from a source domain to more closely reflect the look and feel of a target domain. With no need for paired data, the architecture's built-in capacity to learn domain mappings makes it easier to transfer knowledge between dissimilar domains. By placing a strong emphasis on content and style adaptation, we hope to get beyond the problems of domain shift and distribution mismatch, which will improve generalization and classification accuracy in the target domain. We extensively experiment on benchmark datasets to show the efficacy of the suggested approach. Comparing cross-domain performance to baseline models and other adaption strategies, quantitative evaluation using known measures demonstrates the significant improvement attained. Additionally, qualitative assessments demonstrate the successful alteration of images, demonstrating CycleGAN's competence in adjusting both visual appeal and semantic information. By highlighting the need of simultaneously addressing visual style and content, this study makes a contribution to the field of unsupervised domain adaptation. The outcomes highlight CycleGAN's potential as a formidable domain adaptation tool, opening the door for improved knowledge transfer and performance in real-world settings across several domains

An agile autonomous car driving assistance using hybrid optimization-based kernel support vector convolutional network

In robotic control systems, autonomous car driving is considered a complicated task. The conventional modular techniques necessitate precise localization, planning, and mapping procedures to provide safe driving. These requirements make autonomous car driving tasks computationally ineffective and sensitive to ecological changes. In recent times, the emergence of deep learning-based approaches in autonomous car driving systems provided promising solutions for complex vision task interpretation, localization, and environmental perception. But, they create distribution mismatch issues and require huge perilous interaction information to offer safe driving with the prediction of future collisions. Therefore, to overcome this issue, a novel Modified Kernel Support vector Convolutional based Hybrid Grasshopper Harris hawk (MKSC-HGH) approach is proposed. The proposed MKSC-HGH approach efficiently forecasts the uncertainties of agile autonomous car driving and guides proper navigation details of the track to avoid the possibility of collision with on-road obstacles and fences. This system inputs camera images captured through sensors, car speed, and manual driving data to analyze and predict safety control and probable collision. The prominent feature representing the ability of Modified Convolutional Neural Network (MCNN) and complex boundary encoding process of Kernel Support Vector Machine (KSVM) with hyperparameter tuning operation of Modified Grasshopper Optimization Algorithm based Improved Harris Hawk Optimization (MGOA-IHHO) algorithm made proposed system applicable to predict future states (ie. visual prediction, speed prediction, and collision prediction) more accurately. The efficiency of the proposed MKSC-HGH approach is investigated using evaluation measures namely completion time, average speed, top speed, and laps time. The simulation outcome shows the superiority of the proposed MKSC-HGH approach in predicting agile autonomous car driving over state of art techniques about task performance and sample efficiency.

Designing optimal training sets for genomic prediction using adversarial validation with probit regression

AbstractGenomic selection (GS) is a disruptive methodology that is revolutionizing animal and plant breeding. However, its practical implementation is challenging since many times there is a mismatch in the distribution of the training and testing sets. Adversarial validation is an approach popular in machine learning to detect and address the difference between the training and testing distributions. For this reason, the adversarial validation method in this research was implemented using probit regression to detect the mismatch in distributions and also to select an optimal training set. We evaluated the proposed method with 14 datasets, and the results were benchmarked regarding of using the whole reference population and simple random samples. We found that the proposed method is effective for detecting the mismatch in distributions and outperformed in prediction accuracy by 11.67% (in terms of mean square error) and by 5.35% (in terms of normalized mean square error) when the whole reference population was used as training sets. Also, in general, this outperformed some existing methods for optimal training designs in the context of GS.

Model-Based Offline Reinforcement Learning with Local Misspecification

We present a model-based offline reinforcement learning policy performance lower bound that explicitly captures dynamics model misspecification and distribution mismatch and we propose an empirical algorithm for optimal offline policy selection. Theoretically, we prove a novel safe policy improvement theorem by establishing pessimism approximations to the value function. Our key insight is to jointly consider selecting over dynamics models and policies: as long as a dynamics model can accurately represent the dynamics of the state-action pairs visited by a given policy, it is possible to approximate the value of that particular policy. We analyze our lower bound in the LQR setting and also show competitive performance to previous lower bounds on policy selection across a set of D4RL tasks.

DM²: Decentralized Multi-Agent Reinforcement Learning via Distribution Matching

Current approaches to multi-agent cooperation rely heavily on centralized mechanisms or explicit communication protocols to ensure convergence. This paper studies the problem of distributed multi-agent learning without resorting to centralized components or explicit communication. It examines the use of distribution matching to facilitate the coordination of independent agents. In the proposed scheme, each agent independently minimizes the distribution mismatch to the corresponding component of a target visitation distribution. The theoretical analysis shows that under certain conditions, each agent minimizing its individual distribution mismatch allows the convergence to the joint policy that generated the target distribution. Further, if the target distribution is from a joint policy that optimizes a cooperative task, the optimal policy for a combination of this task reward and the distribution matching reward is the same joint policy. This insight is used to formulate a practical algorithm (DM^2), in which each individual agent matches a target distribution derived from concurrently sampled trajectories from a joint expert policy. Experimental validation on the StarCraft domain shows that combining (1) a task reward, and (2) a distribution matching reward for expert demonstrations for the same task, allows agents to outperform a naive distributed baseline. Additional experiments probe the conditions under which expert demonstrations need to be sampled to obtain the learning benefits.

Discriminable feature enhancement for unsupervised domain adaptation

Unsupervised domain adaptation addresses the problem of knowledge transformation from source domain to target domain, aiming to effectively alleviate data distribution mismatch and data labeling consumption. The issue of data distribution mismatches is widespread in actual agricultural visual tasks. Moreover, it is expensive and time-consuming to construct and label visual image data. For in-field cotton boll, its maturing status can greatly affect the yield and quality. Uneven distribution restrains the performance for maturing status recognition. Therefore, domain adaptation is essential for identifying cross-domain cotton boll maturity. Existing unsupervised domain adaptation methods obtain domain invariant feature for achieving domain alignment. However, the discriminability of features is less considered, which may result in unsatisfactory classification results. In this paper, an unsupervised domain adaptation method called discriminable feature enhancement(DFE-DA) is proposed to identify cross-domain cotton boll maturity. It enables to minimize intra-class distance by maximizing intra-domain density(MID) loss and realizes discriminable feature enhancement. The effectiveness of DFE-DA is verified on in-field cotton boll V2(IFCB-V2) dataset containing 2400 images. The experimental results demonstrate that DFE-DA has an average improvement of 12.8%, 10.3% and 7.6% compared with other methods in three different transfer tasks. Furthermore, the MID loss can cooperate well with other adversarial methods. To verity the robustness of DFE-DA, additional experiments conducted on the public benchmark Office-31 and Office-Home indicates it is comparable to the state-of-the-arts.

Domain embedding transfer for unequal RGB-D image recognition

Most recent unsupervised domain adaptation (UDA) approaches concentrate on single RGB source to single RGB target task. They have to face the real-world scenario, where the source domain can be collected from multiple modalities, e.g., RGB data and depth data. Our work focuses on a more practical and challenging scenario which recognizes RGB images by learning from RGB-D data under the label inequality scenario. We are confronted with three challenges: multiple modalities in the source domain, domain shifting problem and unequal label numbers. To address the aforementioned settings, a novel method, referred to as Domain depth Embedding Transfer (DdET) is proposed, which takes advantage of the depth data in the source domain and handles the domain distribution mismatch under label inequality scenario simultaneously. We conduct comprehensive experiments on five cross domain image classification tasks and observe that DdET can perform favorably against state-of-the-art methods, especially under label inequality scenario.

Translating AI to Clinical Practice: Overcoming Data Shift with Explainability.

To translate artificial intelligence (AI) algorithms into clinical practice requires generalizability of models to real-world data. One of the main obstacles to generalizability is data shift, a data distribution mismatch between model training and real environments. Explainable AI techniques offer tools to detect and mitigate the data shift problem and develop reliable AI for clinical practice. Most medical AI is trained with datasets gathered from limited environments, such as restricted disease populations and center-dependent acquisition conditions. The data shift that commonly exists in the limited training set often causes a significant performance decrease in the deployment environment. To develop a medical application, it is important to detect potential data shift and its impact on clinical translation. During AI training stages, from premodel analysis to in-model and post hoc explanations, explainability can play a key role in detecting model susceptibility to data shift, which is otherwise hidden because the test data have the same biased distribution as the training data. Performance-based model assessments cannot effectively distinguish the model overfitting to training data bias without enriched test sets from external environments. In the absence of such external data, explainability techniques can aid in translating AI to clinical practice as a tool to detect and mitigate potential failures due to data shift. ©RSNA, 2023 Quiz questions for this article are available in the supplemental material.

Spatio‐temporal pattern and driving mechanisms of land use conflicts changes (2010–2018) in the Bohai Rim transition zone

AbstractWith the rapid development of society and economy in the coastal regions, conflicts between ecological protection and human exploitation are inevitable. However, previous studies on the Spatio‐temporal evolution of conflicts in the coastal transition zone are insufficient, and the majority ignore the dependence of data on space. This study constructed a multi‐objective evaluation model from three angles (ecological, agriculture, and construction). It evaluated the dynamics of land use conflicts (LUCs) in the Bohai Rim transition zone for 2010, 2015, and 2018 at 30 m2 spatial resolution using the multi‐factor superposition method. The driving mechanism of LUCs was explored from the perspectives of natural and socio‐economic by applying the spatial econometric method. Major findings of this study indicated that there was a critical Spatio‐temporal mismatch in the spatial distribution of three types of land suitability around the Bohai Sea from 2010 to 2018. On the grid scale, an estimated LUCs area in 2010 was 1.21 × 105 km2, which decreased to 1.20 × 105 km2 in 2018. On the country scale, the conflict index displayed a polarized spatial distribution of “high in the south and low in the north,” and its spatial disparity showed a narrowing trend. Moreover, LUC types showed different change characteristics along the sea‐land direction. It ought to be noticed that the spatial clustering pattern of LUCs resulted from the combined action of various factors. This study's logical framework and outcomes could be utilized for formulating reasonable land use patterns to reduce the LUCs in the coastal transition zone.

Dataset Similarity to Assess Semisupervised Learning Under Distribution Mismatch Between the Labeled and Unlabeled Datasets

Semi-supervised deep learning (SSDL) is a popular strategy to leverage unlabelled data for machine learning when labelled data is not readily available. In real-world scenarios, different unlabelled data sources are usually available, with varying degrees of distribution mismatch regarding the labelled datasets. It begs the question which unlabelled dataset to choose for good SSDL outcomes. ftentimes, semantic heuristics are used to match unlabelled data with labelled data. However, a quantitative and systematic approach to this election problem would be preferable. In this work, we first test the SSDL MixMatch algorithm under various distribution mismatch configurations to study the impact on SSDL accuracy. Then, we propose a quantitative unlabelled dataset selection heuristic based on dataset dissimilarity measures. These are designed to systematically assess how distribution mismatch between the labelled and unlabelled datasets affects MixMatch performance. We refer to our proposed method as deep dataset dissimilarity measures (DeDiMs), designed to compare labelled and unlabelled datasets. They use the feature space of a generic Wide-ResNet, can be applied prior to learning, are quick to evaluate and model agnostic. The strong correlation in our tests between MixMatch accuracy and the proposed DeDiMs suggests that this approach can be a good fit for quantitatively ranking different unlabelled datasets prior to SSDL training.

An Energy-and-Area-Efficient CNN Accelerator for Universal Powers-of-Two Quantization

CNN model computation on edge devices is tightly restricted to the limited resource and power budgets, which motivates the low-bit quantization technology to compress CNN models into 4-bit or lower format to reduce the model size and increase hardware efficiency. Most current low-bit quantization methods use uniform quantization that maps weight and activation values onto evenly-distributed levels, which usually results in accuracy loss due to distribution mismatch. Meanwhile, some non-uniform quantization methods propose specialized representation that can better match various distribution shapes but are usually difficult to be efficiently accelerated on hardware. In order to achieve low-bit quantization with high accuracy and hardware efficiency, this paper proposes Universal Power-of-Two (UPoT), a novel low-bit quantization method that represents values as the addition of multiple power-of-two values selected from a series of subsets. By updating the subset contents, UPoT can provide adaptive quantization levels for various distributions. For each CNN model layer, UPoT automatically searches for the optimized distribution that minimizes the quantization error. Moreover, we design an efficient accelerator system with specifically optimized power-of-two multipliers and requantization units. Evaluations show that the proposed architecture can provide high-performance CNN inference with reduced circuit area and energy, and outperforms several mainstream CNN accelerators with higher ( <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$8\times $ </tex-math></inline-formula> – <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$65\times $ </tex-math></inline-formula> ) area efficiency and ( <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$2\times $ </tex-math></inline-formula> – <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$19\times $ </tex-math></inline-formula> ) energy efficiency. Further experiments of 4/3/2-bit quantization on ResNet18/50, MobileNet_V2 and EfficientNet models show that our UPoT can achieve high model accuracy which greatly outperform other state-of-the-art low-bit quantization methods by 0.3%–6%. The results indicate that our approach provides a highly-efficient accelerator for low-bit CNN model quantization with low hardware overheads and good model accuracy.

Active Incremental Learning for Health State Assessment of Dynamic Systems With Unknown Scenarios

This article is concerned with the data-driven health state assessment task with unknown scenarios. Unknown scenarios are inaccessible in the stage of model training but they can appear unexpectedly during the running stage. A new problem called the within-class distribution mismatch is raised by assuming that unknown scenarios still belong to known classes. To tackle this challenging problem, a novel active incremental learning scheme with a classifier and an out-of-distribution (OOD) detector is proposed. A <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">k</i> -fast incremental support vector data description ( <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">k</i> -FISVDD) model is put forward as the OOD detector to recognize distribution mismatch samples online for label annotation. Specifically, it integrates the clustering algorithm to build local support vector sets, based on which an active query strategy is developed. An incremental learning mechanism is also designed to reduce the labeling cost. Then, the new labeled data can simultaneously refine the classifier and OOD detector. Two cases, including a bearing benchmark dataset and the operation data of a practical deep-sea manned submersible, are studied to demonstrate the effectiveness of the proposed scheme.