Please login to be able to save your searches and receive alerts for new content matching your search criteria.
Domain adaptation is an important subfield of transfer learning, it has been successfully applied in many applications of machine learning. Recently, significant theoretical and algorithmic advances have been achieved in domain adaptation. The theoretical analyses for domain adaptation are based on VC dimension and Rademacher complexity. There are also some covering number-based results, but most of these bounds are based on the results of Rademacher complexity, indirectly given by the relationship between covering number and Rademacher complexity. In this paper, we propose a theoretical analysis framework for domain adaptation, thus the error bound can be derived directly by covering number, which is an effective method for analyzing the generalization error in statistical learning theory. We derive generalization error bound for domain adaptation with a class of loss functions satisfying the assumptions. We also propose a mixup contrastive adversarial network for domain adaptation by introducing a mixup module for enhancing the alignment of the source and target domains during domain transfer, and a contrastive learning module for solving class-level alignment after domain transfer. Experimental results demonstrate the effectiveness of the proposed algorithm and the property of the theoretical results.
Emotion plays a significant role in human daily activities, and it can be effectively recognized from EEG signals. However, individual variability limits the generalization of emotion classifiers across subjects. Domain adaptation (DA) is a reliable method to solve the issue. Due to the nonstationarity of EEG, the inferior-quality source domain data bring negative transfer in DA procedures. To solve this problem, an auto-augmentation joint distribution adaptation (AA-JDA) method and a burden-lightened and source-preferred JDA (BLSP-JDA) approach are proposed in this paper. The methods are based on a novel transfer idea, learning the specific knowledge of the target domain from the samples that are appropriate for transfer, which reduces the difficulty of transfer between two domains. On multiple emotion databases, our model shows state-of-the-art performance.
Automatic checkout (ACO) aims at correctly generating complete shopping lists from checkout images. However, the domain gap between the single product in training data and multiple products in checkout images endows ACO tasks with a major difficulty. Despite remarkable advancements in recent years, resolving the significant domain gap remains challenging. It is possibly because networks trained solely on synthesized images may struggle to generalize well to realistic checkout scenarios. To this end, we propose a decoupled edge guidance network (DEGNet), which integrates synthesized and checkout images via a supervised domain adaptation approach and further learns common domain representations using a domain adapter. Specifically, an edge embedding module is designed for generating edge embedding images to introduce edge information. On this basis, we develop a decoupled feature extractor that takes original images and edge embedding images as input to jointly utilize image information and edge information. Furthermore, a novel proposal divide-and-conquer strategy (PDS) is proposed for the purpose of augmenting high-quality samples. Through experimental evaluation, DEGNet achieves state-of-the-art performance on the retail product checkout (RPC) dataset, with checkout accuracy (cAcc) results of 93.47% and 95.25% in the average mode of faster RCNN and cascade RCNN frameworks, respectively. Codes are available at https://github.com/yourbikun/DEGNet.
Surface electromyography (sEMG)-based gesture recognition can achieve high intra-session performance. However, the inter-session performance of gesture recognition decreases sharply due to the shift in data distribution. Therefore, developing a robust model to minimize the data distribution difference is crucial to improving the user experience. In this work, based on the inter-session gesture recognition task, we propose a novel algorithm called locality preserving and maximum margin criterion (LPMM). The LPMM algorithm integrates three main modules, including domain alignment, pseudo-label selection, and iteration result selection. Domain alignment is designed to preserve the neighborhood structure of the feature and minimize the overlap of different classes. The pseudo-label selection and iteration result selection can avoid the decrease in accuracy caused by mislabeled samples. The proposed algorithm was evaluated on two of the most widely used EMG databases. It achieves a mean accuracy of 98.46% and 71.64%, respectively, which is superior to state-of-the-art domain adaptation methods.
Domain adaptation is a subfield of statistical learning theory that takes into account the shift between the distribution of training and test data, typically known as source and target domains, respectively. In this context, this paper presents an incremental approach to tackle the intricate challenge of unsupervised domain adaptation, where labeled data within the target domain is unavailable. The proposed approach, OTP-DA, endeavors to learn a sequence of joint subspaces from both the source and target domains using Linear Discriminant Analysis (LDA), such that the projected data into these subspaces are domain-invariant and well-separated. Nonetheless, the necessity of labeled data for LDA to derive the projection matrix presents a substantial impediment, given the absence of labels within the target domain in the setting of unsupervised domain adaptation. To circumvent this limitation, we introduce a selective label propagation technique grounded on optimal transport (OTP), to generate pseudo-labels for target data, which serve as surrogates for the unknown labels. We anticipate that the process of inferring labels for target data will be substantially streamlined within the acquired latent subspaces, thereby facilitating a self-training mechanism. Furthermore, our paper provides a rigorous theoretical analysis of OTP-DA, underpinned by the concept of weak domain adaptation learners, thereby elucidating the requisite conditions for the proposed approach to solve the problem of unsupervised domain adaptation efficiently. Experimentation across a spectrum of visual domain adaptation problems suggests that OTP-DA exhibits promising efficacy and robustness, positioning it favorably compared to several state-of-the-art methods.
Automatic seizure detection from Electroencephalography (EEG) is of great importance in aiding the diagnosis and treatment of epilepsy due to the advantages of convenience and economy. Existing seizure detection methods are usually patient-specific, the training and testing are carried out on the same patient, limiting their scalability to other patients. To address this issue, we propose a cross-subject seizure detection method via unsupervised domain adaptation. The proposed method aims to obtain seizure specific information through shallow and deep feature alignments. For shallow feature alignment, we use convolutional neural network (CNN) to extract seizure-related features. The distribution gap of the shallow features between different patients is minimized by multi-kernel maximum mean discrepancies (MK-MMD). For deep feature alignment, adversarial learning is utilized. The feature extractor tries to learn feature representations that try to confuse the domain classifier, making the extracted deep features more generalizable to new patients. The performance of our method is evaluated on the CHB-MIT and Siena databases in epoch-based experiments. Additionally, event-based experiments are also conducted on the CHB-MIT dataset. The results validate the feasibility of our method in diminishing the domain disparities among different patients.
Numerous low-resolution (LR) face images are captured by a growing number of surveillance cameras nowadays. In some particular applications, such as suspect identification, it is required to recognize an LR face image captured by the surveillance camera using only one high-resolution (HR) profile face image on the ID card. This leads to LR face recognition with single sample per person (SSPP), which is more challenging than conventional LR face recognition or SSPP face recognition. To address this tough problem, we propose a Boosted Coupled Marginal Fisher Analysis (CMFA) approach, which unites domain adaptation and coupled mappings. An auxiliary database containing multiple HR and LR samples is introduced to explore more discriminative information, and locality preserving domain adaption (LPDA) is designed to realize good domain adaptation between SSPP training set (target domain) and auxiliary database (source domain). We perform LPDA on HR and LR images in both domains, then in the domain adaptation space we apply CMFA to learn the discriminative coupled mappings for classification. The learned coupled mappings embed knowledge from the auxiliary dataset, thus their discriminative ability is superior. We extensively evaluate the proposed method on FERET, LFW and SCface database, the promising results demonstrate its effectiveness on LR face recognition with SSPP.
With the recent technological advances, surveillance cameras became accessible to the general public and a huge amount of nonstructured data is being gathered. However, extracting value from this data is challenging, especially for tasks that involve human images, such as face recognition and person re-identification. Annotation of this kind of data is a challenging and expensive task. In this work, we propose a domain adaptation workflow to allow CNNs that were trained in one domain to be applied to another domain without the need for annotated target data. Our method uses AlignedReID++ as the baseline, trained using a Triplet loss with batch hard. Domain adaptation is done in an unsupervised manner by clustering unlabeled data to generate pseudo-labels in the target domain. Our results show that domain adaptation really improves the performance of the CNN when applied in the target domain.
Distributions of electroencephalogram (EEG) data vary greatly across different subjects. It is a very important issue how to generalize models across subjects. In this paper, an algorithm is proposed to build high-performance cross-subject motor-imagery brain–computer interfaces (BCIs) for a new subject. First, a novel distance metric is proposed to quantify the joint distribution discrepancy (JDD) between data from different subjects. It gives better evaluations for discrepancies between different distributions than conventional probabilistic metrics. Moreover, it can be extended to design many novel algorithms. Second, a support vector machine combined with JDD (JDMSVM) is proposed for cross-subject classification. For dataset dataIVa, the JDMSVM runs best under 9 out of 15 situations and averagely outperforms counterparts by 10.1%, 9.5%, 3.2% and 1.7%, respectively. For GigaDataset, JDMSVM runs best under 8 of 12 conditions. It averagely outperforms its counterparts by 10.4%, 5.3%, 2.7% and 2.4%, respectively. The experiments demonstrate that the proposed algorithm is effective and competitive for cross-subject BCI.
A strong assumption to derive generalization guarantees in the standard PAC framework is that training (or source) data and test (or target) data are drawn according to the same distribution. Because of the presence of possibly outdated data in the training set, or the use of biased collections, this assumption is often violated in real-world applications leading to different source and target distributions. To go around this problem, a new research area known as Domain Adaptation (DA) has recently been introduced giving rise to many adaptation algorithms and theoretical results in the form of generalization bounds. This paper deals with self-labeling DA whose goal is to iteratively incorporate semi-labeled target data in the learning set to progressively adapt the classifier from the source to the target domain. The contribution of this work is three-fold: First, we provide the minimum and necessary theoretical conditions for a self-labeling DA algorithm to perform an actual domain adaptation. Second, following these theoretical recommendations, we design a new iterative DA algorithm, called GESIDA, able to deal with structured data. This algorithm makes use of the new theory of learning with (ε,γ,τ)-good similarity functions introduced by Balcan et al., which does not require the use of a valid kernel to learn well and allows us to induce sparse models. Finally, we apply our algorithm on a structured image classification task and show that self-labeling domain adaptation is a new original way to deal with scaling and rotation problems.
A transfer learning environment is characterized by not having sufficient labeled training data from the domain of interest (target domain) to build a high-performing machine learner. Transfer learning algorithms use labeled data from an alternate domain (source domain), that is similar to the target domain, to build high-performing learners. The design of a transfer learning algorithm is typically comprised of a domain adaptation step following by a learning step. The domain adaptation step attempts to align the distribution differences between the source domain and the target domain. Then, the aligned data from the domain adaptation step is used in the learning step, which is typically implemented with a traditional machine learning algorithm. Our research studies the impact of the learning step on the performance of various transfer learning algorithms. In our experiment, we use five unique domain adaptation methods coupled with seven different traditional machine learning methods to create 35 different transfer learning algorithms. We perform comparative performance analyses of the 35 transfer learning algorithms, along with the seven stand-alone traditional machine learning methods. This research will aid machine learning practitioners in the algorithm selection process for a transfer learning environment in the absence of reliable validation techniques.
Rolling bearings play an important role in rotating machinery. According to statistics, rolling bearings cause one-third faults of rotating machinery. Once a rolling bearing malfunctions, it may induce maintenance, affect work efficiency, or even cause the entire equipment to malfunction. Therefore, accurately determining the operating status of bearings is of great significance for maintaining the health of the rotating machinery. Most current fault detections of rolling bearing works focus on traditional anomaly detection models which assume the training set to follow the same distribution of test set. This assumption does not hold in fault detection of rolling bearings across different conditions and traditional anomaly detection models may be invalid. This paper introduces domain adaptation anomaly detection (DAAD) in the fault detection of rolling bearings to address this issue. DAAD can adapt anomaly detection across different distributions. The experiments of rolling bearing fault detection under single condition or across different condition show that DAAD is superior to most of the traditional anomaly detection models.
Deep networks have achieved great success in forest fire detection by exploiting visible light images. However, visible light images are susceptible to strong light, smoke, and obstruction interference. The infrared image has high sensitivity to temperature changes of targets, which can alleviate the deficiency of visible light image. Due to the significant distribution shift between visible light and infrared images, directly using the visible light-based pre-trained network for infrared forest fire results in a significant decrease in performance. To resolve this issue, this paper proposes an infrared image forest fire detection system based on domain adaptive learning. We adopt two YOLOv5 frameworks to extract features from visible light images (source domain) and infrared images (target domain). To align the features of the two domains, we construct a novel adaptation learning mechanism based on Kullback–Leibler (KL) loss and feature maximum mean discrepancy (FMMD) loss. We conducted extensive comparative experiments on two publicly available datasets to verify the effectiveness of the proposed model. All experimental results indicate that our proposed domain adaptive learning mechanism effectively improves the performance of infrared forest fire detection.
Human skeleton-based posture anomaly detection has been widely applied in the field of physical education teaching. The existing spatio-temporal graph convolutional networks (ST-GCN) can fully utilize the local and global information of the human skeleton for action recognition, but the entire model requires a large amount of computation and the modeling of high-order relationships between joint points of the human skeleton is insufficient. To this end, this paper proposes a novel domain adaptive hypergraph convolutional network for basketball posture anomaly analysis by exploiting 2D skeleton information. First, we designed an effective hypergraph convolution feature extraction network to improve the high-order dependency modeling. To further improve the performance of the hypergraph convolutional network, we introduce domain adaptive learning technology to supervise the feature extraction learning of the target domain (2D skeleton) through the source domain (3D skeleton). At last, we construct a basketball action teaching analysis dataset for model evaluation. We conducted a large number of comparative experiments on the public dataset NTU RGB+D and our self-built dataset. All the results showed that our proposed hypergraph convolutional model effectively extracts features of 2D human skeletons, and by introducing domain adaptive learning, the performance of basketball anomaly detection is further improved.
Domain adaptation is an important area of research, as it aims to remedy the effects of the domain shift due to differences in the distribution between the source domain used for training and the target domain where prediction takes place. However, methods for characterizing the domain shift across datasets are lacking. In this work, we propose a domain shift metric called SpOT, which stands for spherical optimal transport, by operating on the spherical manifold. We realize our approach with a spherical network, used to obtain features, and an orthogonal projection loss, used to impose orthogonality in the feature space. The resulting spherical features have better inter-class separation and lower intra-class variation compared to features in Euclidean space. This type of feature clustering makes each domain representation more compact and more suitable for further analysis. The domain shift between the datasets is calculated using the optimal transport on the spherical features, which has a sound theoretical basis. Our results are further supported by experiments that show the correlation of SpOT with a new gain of transfer measure across domain adaptation datasets.
Nowadays, with the rapid expansion of social media as a means of quick communication, real-time disaster information is widely disseminated through these platforms. Determining which real-time and multi-modal disaster information can effectively support humanitarian aid has become a major challenge. In this paper, we propose a novel end-to-end model, named GCN-based Multi-modal Domain Adaptation (GMDA), which consists of three essential modules: the GCN-based feature extraction module, the attention-based fusion module and the MMD domain adaptation module. The GCN-based feature extraction module integrates text and image representations through GCNs, while the attention-based fusion module then merges these multi-modal representations using an attention mechanism. Finally, the MMD domain adaptation module is utilized to alleviate the dependence of GMDA on source domain events by computing the maximum mean discrepancy across domains. Our proposed model has been extensively evaluated and has shown superior performance compared to state-of-the-art multi-modal domain adaptation models in terms of F1 score and variance stability.
Recent studies have revealed that deep networks can learn transferable features that generalize well to novel tasks with little or unavailable labeled data for domain adaptation. However, justifying which components of the feature representations can reason about original joint distributions using JMMD within the regime of deep architecture remains unclear. We present a new backpropagation algorithm for JMMD called the Balanced Joint Maximum Mean Discrepancy (B-JMMD) to further reduce the domain discrepancy. B-JMMD achieves the effect of balanced distribution adaptation for deep network architecture, and can be treated as an improved version of JMMD’s backpropagation algorithm. The proposed method leverages the importance of marginal and conditional distributions behind multiple domain-specific layers across domains adaptively to get a good match for the joint distributions in a second-order reproducing kernel Hilbert space. The learning of the proposed method can be performed technically by a special form of stochastic gradient descent, in which the gradient is computed by backpropagation with a strategy of balanced distribution adaptation. Theoretical analysis shows that the proposed B-JMMD is superior to JMMD method. Experiments confirm that our method yields state-of-the-art results with standard datasets.
Top performing algorithms are trained on massive amounts of labeled data. Alternatively, domain adaptation (DA) provides an attractive way to address the few labeled tasks when the labeled data from a different but related domain are available. Motivated by Fisher criterion, we present the novel discriminative regularization term on the latent subspace which incorporates the latent sparse domain transfer (LSDT) model in a unified framework. The key underlying idea is to make samples from one class closer and farther away from different class samples. However, it is nontrivial to design the efficient optimization algorithm. Instead, we construct a convex surrogate relaxation optimization constraint to ease this issue by alternating direction method of multipliers (ADMM) algorithm. Subsequently, we generalize our model in the reproduced kernel Hilbert space (RKHS) for tracking the nonlinear domain shift. Empirical studies demonstrate the performance improvement on the benchmark vision dataset Caltech-4DA.
Machine learning methods, such as neural network (NN) and support vector machine, assume that the training data and the test data are drawn from the same distribution. This assumption may not be satisfied in many real world applications, like long-term financial failure prediction, because the training and test data may each come from different time periods or domains. This paper proposes a novel algorithm known as fuzzy bridged refinement-based domain adaptation to solve the problem of long-term prediction. The algorithm utilizes the fuzzy system and similarity concepts to modify the target instances' labels which were initially predicted by a shift-unaware prediction model. The experiments are performed using three shift-unaware prediction models based on nine different settings including two main situations: (1) there is no labeled instance in the target domain; (2) there are a few labeled instances in the target domain. In these experiments bank failure financial data is used to validate the algorithm. The results demonstrate a significant improvement in the predictive accuracy, particularly in the second situation identified above.
Noninvasive, glucose-monitoring technologies using infrared spectroscopy that have been studied typically require a calibration process that involves blood collection, which renders the methods somewhat invasive. We develop a truly noninvasive, glucose-monitoring technique using mid-infrared spectroscopy that does not require blood collection for calibration by applying domain adaptation (DA) using deep neural networks to train a model that associates blood glucose concentration with mid-infrared spectral data without requiring a training dataset labeled with invasive blood sample measurements. For realizing DA, the distribution of unlabeled spectral data for calibration is considered through adversarial update during training networks for regression to blood glucose concentration. This calibration improved the correlation coefficient between the true blood glucose concentrations and predicted blood glucose concentrations from 0.38 to 0.47. The result indicates that this calibration technique improves prediction accuracy for mid-infrared glucose measurements without any invasively acquired data.