Multi-domain Learning Research Articles

Opportunity recognition in the tension field of knowledge and learning: The case of converging industries

Knowledge-based and learning perspectives alone explain opportunity recognition insufficiently. While knowledge forms the base, learning may help entrepreneurs in novel contexts. Interactions between them add complexity to the analysis of opportunity recognition. Even though many contexts require combinations of knowledge and learning, research on opportunity recognition in these contexts remains scarce. We address this gap with fuzzy-set qualitative comparative analysis (fsQCA) using data from 107 corporate entrepreneurs from converging industries. Converging industries offer a unique context to explore these complexities, requiring entrepreneurs to merge knowledge and learn from new fields. We identify three types with high levels of opportunity recognition: the “broad experienced adapter”, the “specific experienced adapter”, and the “experimenter”. Unlike a simple knowledge-based view suggests, we argue that knowledge is not always necessary. Entrepreneurs compensate for knowledge deficits by combining several learning capabilities. Configurational analysis enriches the theory of how multi-domain knowledge and learning contribute to opportunity recognition.

Acoustic emission-based leakage detection for gas safety valves: Leveraging a multi-domain encoding learning algorithm

Safety valves in gas distribution systems are crucial for preventing excessive pressure buildups. Leakage in these valves compromises system stability, causing equipment damage and environmental pollution. Acoustic emission signal analysis from the valve body is essential for detecting leakage, but these signals are often weak and easily disrupted by noise, complicating detection. This paper proposes a multi-domain encoding learning algorithm to address these challenges. Unlike conventional single-domain methods, it combines two encoding images, the Gramian angular difference field and the Markov transition field, to enhance the comprehensive expression of leakage features. A lightweight convolutional neural network reduces computing resource dependency, and a convolutional block attention mechanism improves feature identification. Experimental results demonstrate that our method detects gas leakage with an accuracy of at least 96.77%. It maintains superior performance even under intense noise interference, offering a promising solution for detecting weak gas leakages in safety valves amidst high background noise.

AnyFace++: Deep Multi-Task, Multi-Domain Learning for Efficient Face AI.

Accurate face detection and subsequent localization of facial landmarks are mandatory steps in many computer vision applications, such as emotion recognition, age estimation, and gender identification. Thanks to advancements in deep learning, numerous facial applications have been developed for human faces. However, most have to employ multiple models to accomplish several tasks simultaneously. As a result, they require more memory usage and increased inference time. Also, less attention is paid to other domains, such as animals and cartoon characters. To address these challenges, we propose an input-agnostic face model, AnyFace++, to perform multiple face-related tasks concurrently. The tasks are face detection and prediction of facial landmarks for human, animal, and cartoon faces, including age estimation, gender classification, and emotion recognition for human faces. We trained the model using deep multi-task, multi-domain learning with a heterogeneous cost function. The experimental results demonstrate that AnyFace++ generates outcomes comparable to cutting-edge models designed for specific domains.

Integrating Domain Knowledge in Multi-Source Classification Tasks

This work presents an extended investigation into multi-domain learning techniques within the context of image and audio classification, with a focus on the latter. In machine learning, collections of data obtained or generated under similar conditions are referred to as domains or data sources. However, the distinct acquisition or generation conditions of these data sources are often overlooked, despite their potential to significantly impact model generalization. Multi-domain learning addresses this challenge by seeking effective methods to train models to perform adequately across all domains seen during the training process. Our study explores a range of model-agnostic multi-domain learning techniques that leverage explicit domain information alongside class labels. Specifically, we delve into three distinct methodologies: a general approach termed Stew, which involves mixing all available data indiscriminately; and two batch domain-regularization methods: Balanced Domains and Loss Sum. These methods are evaluated through several experiments conducted on datasets featuring multiple data sources for audio and image classification tasks. Our findings underscore the importance of considering domain-specific information during the training process. We demonstrate that the application of the Loss Sum method yields notable improvements in model performance (0.79 F1-Score) compared to conventional approaches that blend data from all available domains (0.62 F1-Score). By examining the impact of different multi-domain learning techniques on classification tasks, this study contributes to a deeper understanding of effective strategies for leveraging domain knowledge in machine learning model training.

Interpretable multi-domain meta-transfer learning for few-shot fault diagnosis of rolling bearing under variable working conditions

To address the challenges of accurately diagnosing few-shot fault samples obtained from rolling bearings under variable operating conditions, as well as the issues of black box nature and delayed feedback to guide fault handling in intelligent diagnostic models, this paper proposes an interpretable multi-domain meta-transfer learning method. Firstly, vibration monitoring data of rolling bearings under different operating conditions are collected, and time–frequency domain features are extracted to construct multi-channel one-dimensional temporal samples as inputs. A multi-domain meta-transfer learning framework based on deep convolutional neural networks is then built to perform few-shot learning with multiple tasks under different operating conditions. The output results are reverse-reconstructed through a fusion hierarchical class activation mapping, and the feature maps are assigned different weights to obtain saliency maps corresponding to the inputs, thus improving the interpretability of the output results. Finally, the dataset of bearing vibration data under time-varying rotational speed conditions is used to validate the effectiveness of the proposed method. Experimental results show that the proposed method can achieve accurate fault diagnosis results under variable operating conditions with few-shot samples, and the diagnosis results can be fed back to the input for decision-making, enhancing the interpretability of the model. Compared with other models, it also demonstrates better robustness and accuracy.

Battery Health Estimation Based on Multidomain Transfer Learning

Battery Health Estimation Based on Multidomain Transfer Learning

MultiMatch: Multi-task Learning for Semi-supervised Domain Generalization

Domain generalization (DG) aims at learning a model on source domains to well generalize on the unseen target domain. Although it has achieved great success, most of the existing methods require the label information for all training samples in source domains, which is time-consuming and expensive in the real-world application. In this article, we resort to solving the semi-supervised domain generalization (SSDG) task, where there are a few label information in each source domain. To address the task, we first analyze the theory of multi-domain learning, which highlights that (1) mitigating the impact of domain gap and (2) exploiting all samples to train the model can effectively reduce the generalization error in each source domain so as to improve the quality of pseudo-labels. According to the analysis, we propose MultiMatch, i.e., extending FixMatch to the multi-task learning framework, producing the high-quality pseudo-label for SSDG. To be specific, we consider each training domain as a single task (i.e., local task) and combine all training domains together (i.e., global task) to train an extra task for the unseen test domain. In the multi-task framework, we utilize the independent batch normalization and classifier for each task, which can effectively alleviate the interference from different domains during pseudo-labeling. Also, most of the parameters in the framework are shared, which can be trained by all training samples sufficiently. Moreover, to further boost the pseudo-label accuracy and the model’s generalization, we fuse the predictions from the global task and local task during training and testing, respectively. A series of experiments validate the effectiveness of the proposed method, and it outperforms the existing semi-supervised methods and the SSDG method on several benchmark DG datasets.

Decoupled Training: Return of Frustratingly Easy Multi-Domain Learning

Multi-domain learning (MDL) aims to train a model with minimal average risk across multiple overlapping but non-identical domains. To tackle the challenges of dataset bias and domain domination, numerous MDL approaches have been proposed from the perspectives of seeking commonalities by aligning distributions to reduce domain gap or reserving differences by implementing domain-specific towers, gates, and even experts. MDL models are becoming more and more complex with sophisticated network architectures or loss functions, introducing extra parameters and enlarging computation costs. In this paper, we propose a frustratingly easy and hyperparameter-free multi-domain learning method named Decoupled Training (D-Train). D-Train is a tri-phase general-to-specific training strategy that first pre-trains on all domains to warm up a root model, then post-trains on each domain by splitting into multi-heads, and finally fine-tunes the heads by fixing the backbone, enabling decouple training to achieve domain independence. Despite its extraordinary simplicity and efficiency, D-Train performs remarkably well in extensive evaluations of various datasets from standard benchmarks to applications of satellite imagery and recommender systems.

Multi-Domain Multi-Scale Diffusion Model for Low-Light Image Enhancement

Diffusion models have achieved remarkable progress in low-light image enhancement. However, there remain two practical limitations: (1) existing methods mainly focus on the spatial domain for the diffusion process, while neglecting the essential features in the frequency domain; (2) conventional patch-based sampling strategy inevitably leads to severe checkerboard artifacts due to the uneven overlapping. To address these limitations in one go, we propose a Multi-Domain Multi-Scale (MDMS) diffusion model for low-light image enhancement. In particular, we introduce a spatial-frequency fusion module to seamlessly integrates spatial and frequency information. By leveraging the Multi-Domain Learning (MDL) paradigm, our proposed model is endowed with the capability to adaptively facilitate noise distribution learning, thereby enhancing the quality of the generated images. Meanwhile, we propose a Multi-Scale Sampling (MSS) strategy that follows a divide-ensemble manner by merging the restored patches under different resolutions. Such a multi-scale learning paradigm explicitly derives patch information from different granularities, thus leading to smoother boundaries. Furthermore, we empirically adopt the Bright Channel Prior (BCP) which indicates natural statistical regularity as an additional restoration guidance. Experimental results on LOL and LOLv2 datasets demonstrate that our method achieves state-of-the-art performance for the low-light image enhancement task. Codes are available at https://github.com/Oliiveralien/MDMS.

Multi-Domain Incremental Learning for Face Presentation Attack Detection

Previous face Presentation Attack Detection (PAD) methods aim to improve the effectiveness of cross-domain tasks. However, in real-world scenarios, the original training data of the pre-trained model is not available due to data privacy or other reasons. Under these constraints, general methods for fine-tuning single-target domain data may lose previously learned knowledge, leading to a catastrophic forgetting problem. To address these issues, we propose a multi-domain incremental learning (MDIL) method for PAD, which not only learns knowledge well from the new domain but also maintains the performance of previous domains stably. Specifically, we propose an adaptive domain-specific experts (ADE) framework based on the vision transformer to preserve the discriminability of previous domains. Furthermore, an asymmetric classifier is designed to keep the output distribution of different classifiers consistent, thereby improving the generalization ability. Extensive experiments show that our proposed method achieves state-of-the-art performance compared to prior methods of incremental learning. Excitingly, under more stringent setting conditions, our method approximates or even outperforms the DA/DG-based methods.

Multi-Domain Deep Learning from a Multi-View Perspective for Cross-Border E-commerce Search

Building click-through rate (CTR) and conversion rate (CVR) prediction models for cross-border e-commerce search requires modeling the correlations among multi-domains. Existing multi-domain methods would suffer severely from poor scalability and low efficiency when number of domains increases. To this end, we propose a Domain-Aware Multi-view mOdel (DAMO), which is domain-number-invariant, to effectively leverage cross-domain relations from a multi-view perspective. Specifically, instead of working in the original feature space defined by different domains, DAMO maps everything to a new low-rank multi-view space. To achieve this, DAMO firstly extracts multi-domain features in an explicit feature-interactive manner. These features are parsed to a multi-view extractor to obtain view-invariant and view-specific features. Then a multi-view predictor inputs these two sets of features and outputs view-based predictions. To enforce view-awareness in the predictor, we further propose a lightweight view-attention estimator to dynamically learn the optimal view-specific weights w.r.t. a view-guided loss. Extensive experiments on public and industrial datasets show that compared with state-of-the-art models, our DAMO achieves better performance with lower storage and computational costs. In addition, deploying DAMO to a large-scale cross-border e-commence platform leads to 1.21%, 1.76%, and 1.66% improvements over the existing CGC-based model in the online AB-testing experiment in terms of CTR, CVR, and Gross Merchandises Value, respectively.

Composite Active Learning: Towards Multi-Domain Active Learning with Theoretical Guarantees

Active learning (AL) aims to improve model performance within a fixed labeling budget by choosing the most informative data points to label. Existing AL focuses on the single-domain setting, where all data come from the same domain (e.g., the same dataset). However, many real-world tasks often involve multiple domains. For example, in visual recognition, it is often desirable to train an image classifier that works across different environments (e.g., different backgrounds), where images from each environment constitute one domain. Such a multi-domain AL setting is challenging for prior methods because they (1) ignore the similarity among different domains when assigning labeling budget and (2) fail to handle distribution shift of data across different domains. In this paper, we propose the first general method, dubbed composite active learning (CAL), for multi-domain AL. Our approach explicitly considers the domain-level and instance-level information in the problem; CAL first assigns domain-level budgets according to domain-level importance, which is estimated by optimizing an upper error bound that we develop; with the domain-level budgets, CAL then leverages a certain instance-level query strategy to select samples to label from each domain. Our theoretical analysis shows that our method achieves a better error bound compared to current AL methods. Our empirical results demonstrate that our approach significantly outperforms the state-of-the-art AL methods on both synthetic and real-world multi-domain datasets. Code is available at https://github.com/Wang-ML-Lab/multi-domain-active-learning.

MDFL: Multi-Domain Diffusion-Driven Feature Learning

High-dimensional images, known for their rich semantic information, are widely applied in remote sensing and other fields. The spatial information in these images reflects the object's texture features, while the spectral information reveals the potential spectral representations across different bands. Currently, the understanding of high-dimensional images remains limited to a single-domain perspective with performance degradation. Motivated by the masking texture effect observed in the human visual system, we present a multi-domain diffusion-driven feature learning network (MDFL) , a scheme to redefine the effective information domain that the model really focuses on. This method employs diffusion-based posterior sampling to explicitly consider joint information interactions between the high-dimensional manifold structures in the spectral, spatial, and frequency domains, thereby eliminating the influence of masking texture effects in visual models. Additionally, we introduce a feature reuse mechanism to gather deep and raw features of high-dimensional data. We demonstrate that MDFL significantly improves the feature extraction performance of high-dimensional data, thereby providing a powerful aid for revealing the intrinsic patterns and structures of such data. The experimental results on three multi-modal remote sensing datasets show that MDFL reaches an average overall accuracy of 98.25%, outperforming various state-of-the-art baseline schemes. Code available at https://github.com/LDXDU/MDFL-AAAI-24.

Federated Adaptive Prompt Tuning for Multi-Domain Collaborative Learning

Federated learning (FL) enables multiple clients to collaboratively train a global model without disclosing their data. Previous researches often require training the complete model parameters. However, the emergence of powerful pre-trained models makes it possible to achieve higher performance with fewer learnable parameters in FL. In this paper, we propose a federated adaptive prompt tuning algorithm, FedAPT, for multi-domain collaborative image classification with powerful foundation models, like CLIP. Compared with direct federated prompt tuning, our core idea is to adaptively unlock specific domain knowledge for each test sample in order to provide them with personalized prompts. To implement this idea, we design an adaptive prompt tuning module, which consists of a meta prompt, an adaptive network, and some keys. The server randomly generates a set of keys and assigns a unique key to each client. Then all clients cooperatively train the global adaptive network and meta prompt with the local datasets and the frozen keys. Ultimately, the global aggregation model can assign a personalized prompt to CLIP based on the domain features of each test sample. We perform extensive experiments on two multi-domain image classification datasets across two different settings -- supervised and unsupervised. The results show that FedAPT can achieve better performance with less than 10% of the number of parameters of the fully trained model, and the global model can perform well in diverse client domains simultaneously.

End-to-End Light Field Image Compression with Multi-Domain Feature Learning

Recently, end-to-end light field image compression methods have been explored to improve compression efficiency. However, these methods have difficulty in efficiently utilizing multi-domain features and their correlation, resulting in limited improvement in compression performance. To address this problem, a novel multi-domain feature learning-based light field image compression network (MFLFIC-Net) is proposed to improve compression efficiency. Specifically, an EPI-based angle completion module (E-ACM) is developed to obtain a complete angle feature by fully exploring the angle information with a large disparity contained in the epipolar plane image (EPI) domain. Furthermore, in order to effectively reduce redundant information in the light field image, a spatial-angle joint transform module (SAJTM) is proposed to reduce redundancy by modeling the intrinsic correlation between spatial and angle features. Experimental results demonstrate that MFLFIC-Net achieves superior performance on MS-SSIM and PSNR metrics compared to public state-of-the-art methods.

Federated clustered multi-domain learning for health monitoring

Wearable Internet of Things (WIoT) and Artificial Intelligence (AI) are rapidly emerging technologies for healthcare. These technologies enable seamless data collection and precise analysis toward fast, resource-abundant, and personalized patient care. However, conventional machine learning workflow requires data to be transferred to the remote cloud server, which leads to significant privacy concerns. To tackle this problem, researchers have proposed federated learning, where end-point users collaboratively learn a shared model without sharing local data. However, data heterogeneity, i.e., variations in data distributions within a client (intra-client) or across clients (inter-client), degrades the performance of federated learning. Existing state-of-the-art methods mainly consider inter-client data heterogeneity, whereas intra-client variations have not received much attention. To address intra-client variations in federated learning, we propose a federated clustered multi-domain learning algorithm based on ClusterGAN, multi-domain learning, and graph neural networks. We applied the proposed algorithm to a case study on stress-level prediction, and our proposed algorithm outperforms two state-of-the-art methods by 4.4% in accuracy and 0.06 in the F1 score. In addition, we demonstrate the effectiveness of the proposed algorithm by investigating variants of its different modules.

Multi-Domain Long-Tailed Learning: Challenges, Progress, and Prospects

Multi-Domain Long-Tailed Learning: Challenges, Progress, and Prospects

Virtual Classification: Modulating Domain-Specific Knowledge for Multidomain Crowd Counting.

Multidomain crowd counting aims to learn a general model for multiple diverse datasets. However, deep networks prefer modeling distributions of the dominant domains instead of all domains, which is known as domain bias. In this study, we propose a simple-yet-effective modulating domain-specific knowledge network (MDKNet) to handle the domain bias issue in multidomain crowd counting. MDKNet is achieved by employing the idea of "modulating", enabling deep network balancing and modeling different distributions of diverse datasets with little bias. Specifically, we propose an instance-specific batch normalization (IsBN) module, which serves as a base modulator to refine the information flow to be adaptive to domain distributions. To precisely modulating the domain-specific information, the domain-guided virtual classifier (DVC) is then introduced to learn a domain-separable latent space. This space is employed as an input guidance for the IsBN modulator, such that the mixture distributions of multiple datasets can be well treated. Extensive experiments performed on popular benchmarks, including Shanghai-tech A/B, QNRF, and NWPU validate the superiority of MDKNet in tackling multidomain crowd counting and the effectiveness for multidomain learning. Code is available at https://github.com/csguomy/MDKNet.

5G and Beyond 5G Technologies Enabling Industry 5.0: Network Applications for Robotics

The convergence of cloud and edge computing, along with distributed AI and the latest 5G/6G communications, is revolutionizing collaboration, connectivity, and interaction. These digital advancements pave the way for a new era of AI-powered robots, enabling them to navigate unfamiliar scenarios and adapt in the long term by seamlessly engaging with the digital realm. Consequently, these innovative applications generate diverse, continuous, and rapidly evolving data transmission requirements that traditional network resource management struggles to satisfy in terms of Quality of Service (QoS). In this paper, we take a step beyond focusing solely on network-side traffic engineering efforts. Instead, we explore the potential of application-side traffic shaping within non-public networks to address these demanding transmission needs. Within the framework of optimizing Quality of Experience, we discuss on how the 5G-ERA project focuses on a multi-domain learning process for autonomous robotics using an intent-based networking approach for optimized resource management within the network validated by the use cases within the project enabling the transition from Industry 4.0 to Industry 5.0. Our central aim is to efficiently control and direct data traffic within the confines of the private network, ensuring it aligns with predefined objectives. While this methodology can be applied in various contexts, it is essential to establish precise intentions rooted in domain expertise. This innovative approach serves as a valuable complement to conventional network resource management methods typically employed at the network infrastructure level.

SGDA: Towards 3-D Universal Pulmonary Nodule Detection via Slice Grouped Domain Attention.

Lung cancer is the leading cause of cancer death worldwide. The best solution for lung cancer is to diagnose the pulmonary nodules in the early stage, which is usually accomplished with the aid of thoracic computed tomography (CT). As deep learning thrives, convolutional neural networks (CNNs) have been introduced into pulmonary nodule detection to help doctors in this labor-intensive task and demonstrated to be very effective. However, the current pulmonary nodule detection methods are usually domain-specific, and cannot satisfy the requirement of working in diverse real-world scenarios. To address this issue, we propose a slice grouped domain attention (SGDA) module to enhance the generalization capability of the pulmonary nodule detection networks. This attention module works in the axial, coronal, and sagittal directions. In each direction, we divide the input feature into groups, and for each group, we utilize a universal adapter bank to capture the feature subspaces of the domains spanned by all pulmonary nodule datasets. Then the bank outputs are combined from the perspective of domain to modulate the input group. Extensive experiments demonstrate that SGDA enables substantially better multi-domain pulmonary nodule detection performance compared with the state-of-the-art multi-domain learning methods.