Unknown Domain Research Articles

Biochemical Characterization of a Heat-Resistant κ-Carrageenase Capable of Tolerating High Temperatures up to 100 °C.

κ-Carrageenase plays a crucial role in the high-value utilization of carrageenan. Heat resistance is a key factor in the practical application of κ-carrageenase, as carrageenan exhibits gel-like properties. Previous studies have shown that the C-terminal noncatalytic domains (nonCDs) can affect the thermostability of κ-carrageenases. In this study, we expressed and characterized a κ-carrageenase, MtKC16A, which contains three nonCDs, from Microbulbifer thermotolerans. MtKC16A has the highest activity at 80 °C and pH 7.0. Surprisingly, it exhibits excellent heat resistance, with 71.58% relative activity at 100 °C and still retains over 50% residual activity after incubation at 100 °C for 60 min. Additionally, MtKC16A has been shown to have a dual substrate hydrolysis activity. It can degrade κ-carrageenan to produce highly single Nκ4 and degrade β/κ-carrageenan to produce Nκ2 and desulfated Nκ4 DA-G-DA-G4S, suggesting its potential in producing κ- and β/κ-hybrid oligosaccharides. Furthermore, we found that the unknown function domain (UNFD) in MtKC16A plays the most vital role among the three nonCDs. When this UNFD is truncated, the resulting mutants completely lose their catalytic ability at 100 °C. Finally, by introducing this UNFD to the C-terminal of another κ-carrageenase CaKC16B, we were able to improve its heat resistance at 100 °C.

Identification and characterization of a chondroitinase ABC with a novel carbohydrate-binding module

Chondroitinases play important roles in structural and functional studies of chondroitin sulfates. Carbohydrate-binding module (CBM) is generally considered as an accessory module in carbohydrate-active enzymes, which promotes the association of the appended enzyme with the substrate and potentiates the catalytic activity. However, the role of natural CBM in chondroitinases has not been investigated. Herein, a novel chondroitinase ChABC29So containing an unknown domain with a predicted β-sandwich fold was discovered from Segatella oris. Recombinant ChABC29So showed enzyme activity towards chondroitin sulfates and hyaluronic acid and acted in a random endo-acting manner. The unknown domain exhibited a chondroitin sulfate-binding capacity and was identified as a CBM. Biochemical characterization of ChABC29So and the CBM-truncated enzyme revealed that the CBM enhances the catalytic activity, thermostability, and disaccharide proportion in the final enzymatic products of ChABC29So. These findings demonstrate the role of the natural CBM in a chondroitinase and will guide future modification of chondroitinases.

Distributed consensus and formation control of multi-AUV systems under actuator faults and switching topology

This note presents the distributed consensus and formation control for a group of AUVs, comprising one leader and three followers arranged in a diamond formation. The study addresses significant control challenges, including external disturbances, noise, model uncertainties, actuator faults, stochastic switching topologies, time-varying communication delays, and positional information between agents. Stochastic switching topologies are assumed to follow a Markov chain. To effectively address these challenges and ensuring satisfactory performance levels, two control architectures have been formulated to facilitate the imposition of desired trajectories upon the system states. The initial architecture amalgamates sliding mode control methodology with adaptive algorithms, designed explicitly for tracking missions in the presence of fully functional actuators. In response to potential actuator failures, the second control architecture integrates passive fault-tolerant techniques, significantly enhancing system reliability under such circumstances as well as switching topology. The proposed framework demonstrates its effectiveness in managing the high nonlinearity and coupled dynamics of AUVs. Simulation results validate the efficacy of the developed strategy. It successfully establishes a desired consensus and formation among agents, reduces chattering phenomena, accurately tracks reference trajectories, handles disturbances and uncertainties within an unknown domain, and achieves finite-time convergence.

Chemical fault diagnosis network based on single domain generalization

Recent developments in fault diagnosis have leveraged domain generalization to address the issue of domain shift. Most existing methods focus on learning domain-invariant representations from multiple source domains. However, collecting valuable fault samples from varying operational conditions is challenging, and it is common for available data to originate from a single operational condition. Thus, this paper introduces a Multi-scale generative and adversarial Metric networks (MGAMN) for Chemical Process Fault Diagnosis. To enhance model generalization, a domain generation module was developed to create augmented domains with significant distributional differences from the source domain. The diagnostic task module then extracts domain-invariant features from both the source and augmented domains. A multi-scale generation strategy is established, utilizing multi-scale deep separable convolutions (Dsc) to ensure that the generated samples contain rich state information. Additionally, an adversarial training and metric learning strategy is designed to learn generalized features capable of resisting unknown domain shifts. Extensive diagnostic experiments on the non-isothermal continuous stirred tank reactor (CSTR) and the Tennessee Eastman Process (TEP) chemical datasets validate the effectiveness of the proposed method. Moreover, ablation studies confirm the effectiveness of the proposed modular strategy, demonstrating significant potential for practical applications.

Meta-learning based infrared ship object detection model for generalization to unknown domains

Infrared images exhibit considerable variations in probability distributions, stemming from the utilization of distinct infrared sensors and the influence of diverse environmental conditions. The variations pose great challenges for deep learning models to detect ship objects and adapt to unseen maritime environments. To address the domain shift problem, we propose an end-to-end infrared ship object detection model based on meta-learning neural network to improve domain adaptation for target domain where data is not available at training phase. Different from existing domain generalization methods, the novelty of our model lies in the effective exploitation of meta-learning and domain adaptation, ensuring that the extracted domain-independent features are meaningful and domain-invariant at the semantic level. Firstly, a double gradient-based meta-learning algorithm is designed to solve the common optimal descent direction between different domains through two gradient updates in the inner and outer loops. The algorithm enables extraction of domain-invariant features from the pseudo-source and pseudo-target domain data. Secondly, a domain discriminator with dynamic-weighted gradient reversal layer (DWGRL) is designed to accurately classify domain-invariant features and provide additional global supervision information. Finally, a multi-scale feature aggregation method is proposed to improve the extraction of multi-scale domain-invariant features. It can effectively fuse local features at different scales and global features of targets. Extensive experimental results conducted in real nighttime water surface scenes demonstrate that the proposed model achieves very high detection accuracy on target domain data, even no target domain data was used during the training phase. Compared to the existing methods, our method not only improves the detection accuracy of infrared ships by 18%, but also exhibits the smallest standard deviation with a value of 0.93, indicating its superior generalization performance.

X-ray structure and mutagenesis analyses of Clostridioides difficile endolysin Ecd09610 glucosaminidase domain

Clostridioides difficile endolysin (Ecd09610) consists of an unknown domain at its N terminus, followed by two catalytic domains, a glucosaminidase domain and endopeptidase domain. X-ray structure and mutagenesis analyses of the Ecd09610 catalytic domain with glucosaminidase activity (Ecd09610CD53) were performed. Ecd09610CD53 was found to possess an α-bundle-like structure with nine helices, which is well conserved among GH73 family enzymes. The mutagenesis analysis based on X-ray structures showed that Glu405 and Asn470 were essential for enzymatic activity. Ecd09610CD53 may adopt a neighboring-group mechanism for a catalytic reaction in which Glu405 acted as an acid/base catalyst and Asn470 helped to stabilize the oxazolinium ion intermediate. Structural comparisons with the newly identified Clostridium perfringens autolysin catalytic domain (AcpCD) in the P1 form and a zymography analysis demonstrated that AcpCD was 15-fold more active than Ecd09610CD53. The strength of the glucosaminidase activity of the GH73 family appears to be dependent on the depth of the substrate-binding groove.

SupCon-MPL-DP: Supervised Contrastive Learning with Meta Pseudo Labels for Deepfake Image Detection

Recently, there has been considerable research on deepfake detection. However, most existing methods face challenges in adapting to the advancements in new generative models within unknown domains. In addition, the emergence of new generative models capable of producing and editing high-quality images, such as diffusion, consistency, and LCM, poses a challenge for traditional deepfake training models. These advancements highlight the need for adapting and evolving existing deepfake detection techniques to effectively counter the threats posed by sophisticated image manipulation technologies. In this paper, our objective is to detect deepfake videos in unknown domains using unlabeled data. Specifically, our proposed approach employs Meta Pseudo Labels (MPL) with supervised contrastive learning, so-called SupCon-MPL, allowing the model to be trained on unlabeled images. MPL involves the simultaneous training of both a teacher model and a student model, where the teacher model generates pseudo labels utilized to train the student model. This method aims to enhance the adaptability and robustness of deepfake detection systems against emerging unknown domains. Supervised contrastive learning utilizes labels to compare samples within similar classes more intensively, while encouraging greater distinction from samples in dissimilar classes. This facilitates the learning of features in a diverse set of deepfake images by the model, consequently contributing to the performance of deepfake detection in unknown domains. When utilizing the ResNet50 model as the backbone, SupCon-MPL exhibited an improvement of 1.58% in accuracy compared with traditional MPL in known domain detection. Moreover, in the same generation of unknown domain detection, there was a 1.32% accuracy enhancement, while in the detection of post-generation unknown domains, there was an 8.74% increase in accuracy.

Inhibition of adenylyl cyclase by GTPase-deficient Gαi is mechanistically different from that mediated by receptor-activated Gαi

Signal transduction through G protein-coupled receptors (GPCRs) has been a major focus in cell biology for decades. Numerous disorders are associated with GPCRs that utilize Gi proteins to inhibit adenylyl cyclase (AC) as well as regulate other effectors. Several early studies have successfully defined the AC-interacting domains of several members of Gαi by measuring the loss of activity upon homologous replacements of putative regions of constitutive active Gαi mutants. However, whether such findings can indeed be translated into the context of a receptor-activated Gαi have not been rigorously verified. To address this issue, an array of known and new chimeric mutations was introduced into GTPase-deficient Q204L (QL) and R178C (RC) mutants of Gαi1, followed by examinations on their ability to inhibit AC. Surprisingly, most chimeras failed to abolish the constitutive activity brought on by the QL mutation, while some were able to eliminate the inhibitory activity of RC mutants. Receptor-mediated inhibition of AC was similarly observed in the same chimeric constructs harbouring the pertussis toxin (PTX)-resistant C351I mutation. Moreover, RC-bearing loss-of-function chimeras appeared to be hyper-deactivated by endogenous RGS protein. Molecular docking revealed a potential interaction between AC and the α3/β5 loop of Gαi1. Subsequent cAMP assays support a cooperative action of the α3/β5 loop, the α4 helix, and the α4/β6 loop in mediating AC inhibition by Gαi1-i3. Our results unveiled a notable functional divergence between constitutively active mutants and receptor-activated Gαi1 to inhibit AC, and identified a previously unknown AC-interacting domain of Gαi subunits. These results collectively provide valuable insights on the mechanism of AC inhibition in the cellular environment.

AdvST: Revisiting Data Augmentations for Single Domain Generalization

Single domain generalization (SDG) aims to train a robust model against unknown target domain shifts using data from a single source domain. Data augmentation has been proven an effective approach to SDG. However, the utility of standard augmentations, such as translate, or invert, has not been fully exploited in SDG; practically, these augmentations are used as a part of a data preprocessing procedure. Although it is intuitive to use many such augmentations to boost the robustness of a model to out-of-distribution domain shifts, we lack a principled approach to harvest the benefit brought from multiple these augmentations. Here, we conceptualize standard data augmentations with learnable parameters as semantics transformations that can manipulate certain semantics of a sample, such as the geometry or color of an image. Then, we propose Adversarial learning with Semantics Transformations (AdvST) that augments the source domain data with semantics transformations and learns a robust model with the augmented data. We theoretically show that AdvST essentially optimizes a distributionally robust optimization objective defined on a set of semantics distributions induced by the parameters of semantics transformations. We demonstrate that AdvST can produce samples that expand the coverage on target domain data. Compared with the state-of-the-art methods, AdvST, despite being a simple method, is surprisingly competitive and achieves the best average SDG performance on the Digits, PACS, and DomainNet datasets. Our code is available at https://github.com/gtzheng/AdvST.

LogFormer: A Pre-train and Tuning Pipeline for Log Anomaly Detection

Log anomaly detection is a key component in the field of artificial intelligence for IT operations (AIOps). Considering log data of variant domains, retraining the whole network for unknown domains is inefficient in real industrial scenarios. However, previous deep models merely focused on extracting the semantics of log sequences in the same domain, leading to poor generalization on multi-domain logs. To alleviate this issue, we propose a unified Transformer-based framework for Log anomaly detection (LogFormer) to improve the generalization ability across different domains, where we establish a two-stage process including the pre-training and adapter-based tuning stage. Specifically, our model is first pre-trained on the source domain to obtain shared semantic knowledge of log data. Then, we transfer such knowledge to the target domain via shared parameters. Besides, the Log-Attention module is proposed to supplement the information ignored by the log-paring. The proposed method is evaluated on three public datasets and one real-world dataset. Experimental results on multiple benchmarks demonstrate the effectiveness of our LogFormer with fewer trainable parameters and lower training costs.

Stochastic neuro-fuzzy system implemented in memristor crossbar arrays.

Neuro-symbolic artificial intelligence has garnered considerable attention amid increasing industry demands for high-performance neural networks that are interpretable and adaptable to previously unknown problem domains with minimal reconfiguration. However, implementing neuro-symbolic hardware is challenging due to the complexity in symbolic knowledge representation and calculation. We experimentally demonstrated a memristor-based neuro-fuzzy hardware based on TiN/TaOx/HfOx/TiN chips that is superior to its silicon-based counterpart in terms of throughput and energy efficiency by using array topological structure for knowledge representation and physical laws for computing. Intrinsic memristor variability is fully exploited to increase robustness in knowledge representation. A hybrid in situ training strategy is proposed for error minimizing in training. The hardware adapts easier to a previously unknown environment, achieving ~6.6 times faster convergence and ~6 times lower error than deep learning. The hardware energy efficiency is over two orders of magnitude greater than field-programmable gate arrays. This research greatly extends the capability of memristor-based neuromorphic computing systems in artificial intelligence.

Leveraging deep learning with progressive growing GAN and ensemble smoother with multiple data assimilation for inverse modeling

The incorporation of hard data in geostatistical modeling is crucial for enhancing the accuracy of interpolating or stochastically estimating subsurface spatial features. The hard data at specified points in the model domain serve as a guide in optimizing the unknown parameters to follow the patterns of the hard data. Recently, a novel approach to solving hydrogeologic/reservoir modeling problems has emerged by using deep generative models, specifically generative adversarial networks (GANs), to generate realistic and diverse images of channelized aquifers. This subsequently can be coupled with inverse models to solve parameter estimation problems. This study focused on using an improved GAN, called a progressive growing generative adversarial network (PGGAN), conditioned with hard data to perform parameter estimation of complex facies models by coupling an ensemble smoother with multiple data assimilation (ES-MDA). First, the PGGAN was trained to an image with 128 × 128 resolution. The trained PGGAN was used to generate hydraulic conductivity fields when fed an ensemble of latent variables and hard data. The ES-MDA then was used to update the latent variable with the help of hydraulic head data obtained from the groundwater model. The approach was tested on synthetic hydraulic conductivity data. Results show that this approach was able to perform efficient estimation of an unknown facies model domain. Additionally, the proposed method was applied to a different test case of a facies model exhibiting different statistical characteristics. The results were satisfactory, demonstrating that the method is not constrained to the particular hydraulic conductivity fields introduced in the generator’s training.

Curriculum learning-based domain generalization for cross-domain fault diagnosis with category shift

Intelligent fault diagnosis has witnessed significant advancements in the preceding years. Domain generalization-based methods can effectively alleviate the domain shift problem and be employ for fault diagnosis in unknown domains. Apart from the problem of domain shift, another challenge arises from the incomplete label space of each source domain due to the difficulty of data acquisition. Category shift can have a significant impact on the subsequent application of intelligent algorithms. To confront this more challenging and practical problem, we begin by formulating the setting of domain generalization with category shift. This paper proposes a Curriculum Learning-based Domain Generalization method (CLDG) to tackle with the intricate problem. The basic network consists of a feature extractor, a mixup-based reciprocal point learning classifier for tackling the category shift between the source and target domains, and a conditional domain discriminator for addressing the domain shift. In addition, we construct a curriculum learning strategy that uses the knowledge of categories with high observation degree to assist in extracting domain invariant features of lower ones, dealing with the category shift between the source domains and improving the generalization ability of the categorical information. Extensive experimental results on two datasets provide evidence for the effectiveness and superiority of the proposed algorithm in classifying known and missing classes in each source domain, as well as identifying unobserved failure modes in unknown target domains.

Rescaling bootstrap variance estimation technique under dual frame surveys with unknown domain sizes

Dual frame (DF) surveys are a special case of multiple frame (MF) surveys considering two frames covering the entire population. Dual frame surveys are applicable in those situations, where, one frame may cover the entire population but is very expensive to sample; so an alternate frame may be available that does not cover the entire population but is easily available. Unbiased variance estimation in dual frame surveys can be difficult and complicated than corresponding estimators under single frame surveys. Again, the variance of dual frame estimator involves population variances of the individual domains which are generally unknown. Due to this reason, obtaining an unbiased estimate of the variance of the dual frame estimator is quite complex in the case of dual frame surveys. In this article, we propose a Post-stratified Rescaling Bootstrap with Unknown Domain size (PstRBUD) method for variance estimation of the dual frame estimator of population total. The proposed rescaled bootstrap method was compared to that of standard bootstrap methods in simulation analysis. The proposed PstRBUD method provides an unbiased estimation of the variance of the dual frame estimator of population total, according to simulation results.

A Simple Unsupervised Knowledge-Free Domain Adaptation for Speaker Recognition

Despite the great success of speaker recognition models based on deep neural networks, deploying a pre-trained model in real-world scenarios often leads to significant performance degradation due to the domain mismatch between training and testing conditions. Various adaptation methods have been developed to address this issue by modifying either the front-end embedding network or the back-end scoring model. However, existing methods typically rely on knowledge of the network, scoring model, or even the source data. In this study, we introduce a knowledge-free adaptation approach that only necessitates the unlabeled target data. Our core concept is based on the assumption that domain mismatch primarily stems from distributional distortion in the embedding space, such as shifting, rotation, and scaling while maintaining inter-speaker discrimination for data from unknown domains. Building on this assumption, we propose clustering LDA (C-LDA), a full-rank linear discriminant analysis (LDA) based on agglomerative hierarchical clustering (AHC) to compensate for this distortion. This approach does not need any human labels and does not rely on any knowledge of the model in the source domain, making it suitable for real-world applications. Theoretical analysis indicates that with cosine scoring, C-LDA is capable of eliminating distributional distortion related to global shift and within-speaker covariance rotation and scaling. Surprisingly, our experiments demonstrated that this simple approach can outperform more complex methods that require full or partial knowledge, including front-end approaches such as fine-tuning and distribution alignment, and back-end approaches such as unsupervised probabilistic linear discriminant analysis (PLDA) adaptation. Additional experiments demonstrated that C-LDA is insensitive to hyperparameters and works well in both multi-domain and single-domain adaptation scenarios.

Abstract A069: Study of MUC16/ERBB axis in pancreatic cancer

Abstract Background: Pancreatic cancer is the third leading cause of cancer-related deaths due to its poor prognosis and highly malignant nature. The altered expression of glycans and glycoproteins is associated with malignancy. MUC16 (also known as CA125) is a member of mucin family of glycoproteins known to line epithelial cells for protection and lubrication and is overexpressed in several cancer types, including pancreatic cancer, often linked with enhanced disease progression, metastasis, and reduced patient survival. Significance: Circulating levels of MUC16/CA125 has been used as a biomarker to monitor adenocarcinoma for decades. However, the functional role of CA125/MUC16 is yet to be elucidated due to a lack of studies on its large, heavily glycosylated structure and unknown functional domains. We present evidence that in addition to serving as a biomarker, circulating MUC16 has pro-tumorigenic activities. Hypothesis: Based on recent findings, we hypothesize that isoforms of MUC16 have high avidity for ErbB receptor and integrin complexes, with concomitant activation of downstream signaling cascades that include Akt and FAK, respectively. These result in the increased tumorigenic potential of pancreatic cancer cells in an autocrine manner. Experimental Design: There are four members of ErbB receptors; where receptor dimerization is a function of ligand binding and results in distinct downstream oncogenic signaling cascades. We evaluated the binding of MUC16 to different ErbB receptors, and its capacity to activate downstream signaling pathways using pancreatic cell line models that have different combinations of knock out/down of each ErbB receptor. We evaluated the global changes in the phosphorylation of oncogenic signaling molecules due to MUC16 binding using Reverse Phase Protein Array (RPPA). Results: We found that MUC16 activates Erbb4 (Y1162) along with other members of ErbB receptor family, where glycosylation of MUC16 is crucial for its biological activity. MUC16 enhanced oncogenic signaling through the phosphorylation of various downstream molecules and activated PI3K, RTK, MAPK, and DNA repair pathways. Our future directions include elucidating the molecular mechanism by which MUC16 activates the oncogenic signaling cascades via binding to ErbB receptors using our pancreatic cell line-based models. Conclusions: Understanding the biological activity of MUC16 and its cleaved product will further help in investigating the role of MUC16 in tumor microenvironment and at distant organ sites to influence the metastatic niche. Citation Format: Amina Baniya, Michael A. (Tony) Hollingsworth. Study of MUC16/ERBB axis in pancreatic cancer [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Pancreatic Cancer; 2023 Sep 27-30; Boston, Massachusetts. Philadelphia (PA): AACR; Cancer Res 2024;84(2 Suppl):Abstract nr A069.

A domain adaptation approach to damage classification with an application to bridge monitoring

Data-driven machine-learning algorithms generally suffer from a lack of labelled health-state data, mainly those referring to damage conditions. To address such an issue, population-based structural health monitoring seeks to enrich the original dataset by transferring knowledge from a population of monitored structures. Within this context, this paper presents a transfer learning approach, based on domain adaptation, to leverage information from completely-labelled bridge structure data to accurately predict new instances of an unknown target domain. Since intrinsic structural differences may cause distribution shifts, domain adaptation attempts to minimise the distance between the domains and to learn a mapping within a shared feature space. Specifically, the methodology involves the long-term acquisition of natural frequencies from several structural scenarios. Such damage-sensitive features are then aligned via domain adaptation so that a machine-learning algorithm can effectively utilise the labelled source domain data and generalise well to the unlabelled target-domain data. The described procedure is applied to two case studies, including the Z24 and the S101 benchmark bridges and their finite element models, respectively. The results demonstrate the successful exchange of health-state labels to identify the damage class within a population of bridges equipped with SHM systems, showing potential to reduce computational efforts and to deal with scarce or poor data sets in application to bridge network monitoring.

Functional Analysis of KAP1/TRIM28 Requirements for HIV-1 Transcription Activation.

HIV-1 latency maintenance and reactivation are regulated by several viral and host factors. One such factor is Krüppel-associated box (KRAB)-associated protein 1 (KAP1: also named TRIM28 or TIF1β). While initial studies have revealed KAP1 to be a positive regulator of latency reversal in transformed and primary CD4+ T cells, subsequent studies have proposed KAP1 to be a repressor required for latency maintenance. Given this discrepancy, in this study, we re-examine KAP1 transcription regulatory functions using a chemical genetics strategy to acutely deplete KAP1 expression to avoid the accumulation of indirect effects. Notably, KAP1 acute loss partially decreased HIV-1 promoter activity in response to activating signals, a function that can be restored upon complementation with exogenous KAP1, thus revealing that KAP1-mediated activation is on target. By combining comprehensive KAP1 domain deletion and mutagenesis in a cell-based reporter assay, we genetically defined the RING finger domain and an Intrinsically Disordered Region as key activating features. Together, our study solidifies the notion that KAP1 activates HIV-1 transcription by exploiting its multi-domain protein arrangement via previously unknown domains and functions.

Zero-Shot Attribute Consistent Model for Bearing Fault Diagnosis Under Unknown Domain

Zero-Shot Attribute Consistent Model for Bearing Fault Diagnosis Under Unknown Domain

Camera-Aware Recurrent Learning and Earth Mover’s Test-Time Adaption for Generalizable Person Re-Identification

Domain generalization in person re-identification (ReID) aims to design a generalizable model, which is trained under the supervision of a set of labeled source domains and can be directly deployed on unknown domains. Existing approaches simply treat each identity as a distinct class and ignore the differences among cameras. We argue that the camera information is crucial for learning discriminative representations, as people’s behavior usually varies between cameras. In this paper, we present Multi-Centroid Memory (MCM) to capture different camera information for each identity and Soft Triple Hard (ST-Hard) loss to align the information of the same identity across cameras. Furthermore, in contrast to the traditional approaches of training a single model using a parallel training mechanism, we propose the Recurrent Implicit Lifelong Learning (RILL) that feeds the source domains into the model in a continuous loop to train an expert for each domain. To make each expert further generalized to other source domains, during the training on the current domain, RILL adopts a style replay-based method to simulate the training of the previous domain, encouraging each domain’s expert to extract generalizable features. We also present Earth Mover’s Test-time Adaption (EMTA) to be used in conjunction with RILL, which enables source domains that are more similar to the test domain to play a more significant role in the test. This is achieved by our proposed Earth Mover’s Similarity (EMS), which helps model the similarities between the source and test domains. Extensive experiments on two evaluation protocols fully demonstrate our framework’s generalization and competitiveness.