Face Recognition via Domain Adaptation and Manifold Distance Metric Learning

Joint metric and feature representation learning for unsupervised domain adaptation

In this paper, an enhanced discriminative feature learning (EDFL) method is proposed to address single sample per person (SSPP) face recognition. With a separate auxiliary dataset, EDFL integrates Fisher discriminative learning and domain adaptation into a unified framework. The separate auxiliary dataset and the gallery/probe dataset are from two different domains (named source and target domains respectively) and have different data distributions. EDFL is modeled to transfer the discriminative knowledge learned from the source domain to the target domain for classification. Since the gallery set with SSPP contains scarce number of samples, it is hard to accurately represent the data distribution of the target domain, which hinders the adaptation effect. To overcome this problem, the generalized domain adaption (GDA) method is proposed to realize good overall domain adaptation when one domain contains limited samples. GDA considers the both global and local domain adaptation effect at the same time. Further, to guarantee that the learned domain adaptation components are optimal for discriminative learning, the domain adaptation and Fisher discriminant model learning are unified into a single framework and an efficient algorithm is designed to optimize them. The effectiveness of the proposed approach is demonstrated by extensive evaluation and comparison with some state-of-the-art methods.

Handwritten character recognition systems suffers from different training and testing sets distributions. In this paper, we propose a two-step domain adaptive multiple kernel learning algorithm, which learns a kernel function based on several kernels in the first step, and trains a target classifier by applying the learned kernel in the second step. Our method can be employed both in semi-supervised and unsupervised domain adaptation cases, while most of the previous domain adaptation methods work only in semi-supervised case. Experiments on adaptation to different databases in this field reveal the superiority of this algorithm in comparison with other adaptation algorithms.

Domain adaptation aims to diminish the discrepancy between the source and target domains and enhance the classification ability for the target samples by using well-labeled source domain data. However, most existing methods concentrate on learning domain invariant features for cross-domain tasks, but ignore the correlation and discriminative information between different domains. If the learned features from the source and target domains are not correlated, the adaptability of domain adaptation methods will be greatly degraded. To make up for this deficiency, we propose a novel domain adaptation approach, referred to as Canonical Correlation Discriminative Learning (CCDL) for domain adaptation. By introducing a novel correlation representation, CCDL maximizes the correlations of the learned features from the two domains as much as possible. Specifically, CCDL learns a latent feature representation to reduce the difference by jointly adapting the marginal and conditional distributions between the source and target domains, and simultaneously maximizes the inter-class distance and minimizes the intra-class scatter. The experiments certify that CCDL is superior to several state-of-the-art methods on four visual benchmark databases.

Domain invariant representation learning is one of the domain adaptation techniques, which aims at learning a representation that is not affected by a possible shift in the data distribution between training and test sets. Traditional domain adaptation methods proposed in the literature focus on first learning a domain invariant transformation from the input space to a new representation space and then training a classifier in the new space. Alternatively, in this paper, we propose an adversarial domain adaptation technique that combines representation learning, domain adaptation and classifier learning in a single training process. Moreover, the proposed method performs the domain adaptation by using the Wasserstein metrics to minimize the domain discrepancy. We applied the proposed method on a hyperspectral image classification problem and the results obtained show the effectiveness of the method.

Face recognition systems are in great demand for domestic and commercial applications. A novel feature extraction approach is proposed based on TanTrigg Lower Edge Directional Patterns for robust face recognition. Histogram of Orientated Gradients is used to detect faces and the facial landmarks are localized using Ensemble of Regression Trees. The detected face is rotated based on facial landmarks using affine transformation followed by cropping and resizing. TanTrigg preprocessor is used to convert the aligned face region into an illumination invariant region for better feature extraction. Eight directional Kirsch compass masks are convolved with the preprocessed face image. Feature descriptor is extracted by dividing the TTLEDP image into several sub-regions and concatenating the histograms of all the sub-regions. Chi-square distance metric is used to match faces from the trained feature space. The experimental results prove that the proposed TTLEDP feature descriptor has better recognition rate than existing methods, overcoming the challenges like varying illumination and noise

Unsupervised domain adaptation techniques have been successful for a wide range of problems where supervised labels are limited. The task is to classify an unlabeled “target” dataset by leveraging a labeled “source” dataset that comes from a slightly similar distribution. We propose metric-based adversarial discriminative domain adaptation (M-ADDA) which performs two main steps. First, it uses a metric learning approach to train the source model on the source dataset by optimizing the triplet loss function. This results in clusters where embeddings of the same label are close to each other and those with different labels are far from one another. Next, it uses the adversarial approach (as that used in ADDA (Tzeng et al. Adversarial discriminative domain adaptation, 2017, [36])) to make the extracted features from the source and target datasets indistinguishable. Simultaneously, we optimize a novel loss function that encourages the target dataset’s embeddings to form clusters. While ADDA and M-ADDA use similar architectures, we show that M-ADDA performs significantly better on the digits adaptation datasets of MNIST and USPS. This suggests that using metric learning for domain adaptation can lead to large improvements in classification accuracy for the domain adaptation task. The code is available at https://github.com/IssamLaradji/M-ADDA.

Sparse feature space representation: A unified framework for semi-supervised and domain adaptation learning

Due to the individual differences and nonstationary of EEG signals, it is difficult to classify EEG emotions with traditional machine methods, which assume that the training and testing set come from the same data distribution, but this assumption is usually not true in the EEG field, therefore the accuracy of emotion recognition is very poor. In this paper, a Single-Source Domain Adaptive Few-Shot Learning Networks (SDA-FSL) was proposed for cross-subject EEG emotion recognition. This is the first time that domain adaptation method with few-shot learning has been used in the field of EEG emotion recognition. A CBAM-based feature mapping module was designed to extract the common features of the two domains, and the domain adaptation module was used to align the data distribution of two domains. In addition, Prototypical Networks with instance-attention mechanism is introduced to preserve domain-specific information. The proposed method was evaluated on DEAP and SEED datasets in within-dataset and cross-dataset experiments under various N-way k-shot settings. Experimental results show that the performance of SDA-FSL outperforms other comparison methods and has superior generalization performance on cross-dataset experiments.

Unsupervised domain adaptation (UDA) aims to transfer knowledge learned from a fully-labeled source domain to a different unlabeled target domain. Most existing UDA methods learn domain-invariant feature representations by minimizing feature distances across domains. In this work, we build upon contrastive self-supervised learning to align features so as to reduce the domain discrepancy between training and testing sets. Exploring the same set of categories shared by both domains, we introduce a simple yet effective framework CDCL, for domain alignment. In particular, given an anchor image from one domain, we minimize its distances to cross-domain samples from the same class relative to those from different categories. Since target labels are unavailable, we use a clustering-based approach with carefully initialized centers to produce pseudo labels. In addition, we demonstrate that CDCL is a general framework and can be adapted to the data-free setting, where the source data are unavailable during training, with minimal modification. We conduct experiments on two widely used domain adaptation benchmarks, i.e., Office-31 and VisDA-2017, for image classification tasks, and demonstrate that CDCL achieves state-of-the-art performance on both datasets.

With the enhancement of air-based and space-based perception capabilities, space-aeronautics incorporation and integration is growing in importance. Full domain awareness is crucial for integrated perception systems, in which domain adaptation is one of the key problems in improving the performance of cross-domain perception. Deep learning is currently an advanced technique for complex inverse synthetic aperture radar (ISAR) object recognition. However, the training procedure needs many annotated samples, which is insufficient for certain targets, such as aircraft. Few-shot learning provides a new approach to solving the above problem by transferring useful knowledge from other domains, such as optical images from satellites. Nevertheless, it fails to fully consider the domain shift between the source and target domains, generally neglecting the transferability of training samples in the learning process. Consequently, it produces suboptimal recognition accuracy. To address the composite problems mentioned above, we propose a domain adaptive few-shot learning method from satellites to an ISAR called S2I-DAFSL for aircraft recognition tasks. Furthermore, unlike conventional domain adaptation methods that directly align the distributions, the attention transferred importance-weighting network (ATIN) is proposed to improve the transferability in the domain adaptation procedure. Compared with state-of-the-art methods, it shows that the proposed method achieves better performance, increasing the accuracy and effectiveness of classification, which is more suitable for cross-domain few-shot ISAR aircraft recognition tasks.

Thanks to digitization of industrial assets in fleets, the ambitious goal of transferring fault diagnosis models from one machine to the other has raised great interest. Solving these domain adaptive transfer learning tasks has the potential to save large efforts on manually labeling data and modifying models for new machines in the same fleet. Although data-driven methods have shown great potential in fault diagnosis applications, their ability to generalize on new machines and new working conditions are limited because of their tendency to overfit to the training set in reality. One promising solution to this problem is to use domain adaptation techniques. It aims to improve model performance on the target new machine. Inspired by its successful implementation in computer vision, we introduced Domain-Adversarial Neural Networks (DANN) to our context, along with two other popular methods existing in previous fault diagnosis research. We then carefully justify the applicability of these methods in realistic fault diagnosis settings, and offer a unified experimental protocol for a fair comparison between domain adaptation methods for fault diagnosis problems.

Domain Adaptation (DA) or Transfer Learning (TL) makes use of the already available source domain information for training the target domain classifier. Traditional ML algorithms require abundant amount of labeled data for training the model, and also they assume that both training and testing data follow similar distributions. However, in a real-world scenario, this does not always work. The scarcity of labeled data in the target domain is a big issue. Also, the source and the target domains have distinct data distributions. So, lessening the gap between the distributions of the two domains is very important so that a model that is trained using source domain information can be deployed to classify the target domain information efficiently. The already existing domain adaptation technique tries to reduce this distribution interval statistically and geometrically to an extent. Nevertheless, it requires some important components such as Laplacian regularization and maximizing source domain variance. Hence, we propose a Modified Joint Geometrical and Statistical Alignment (MJGSA) approach for Low-Resolution Face Recognition that enhances the previous transfer learning methods by incorporating all the necessary objectives that are useful for diminishing the distribution gap between the domains. Rigorous experiments on several real-world datasets verify that our proposed MJGSA approach surpasses other state-of-the-art existing methods.

Domain adaptive twin support vector machine learning using privileged information

Face recognition provides a challenging issue in the domain of analyzing images. In this paper a novel approach for face recognition using hybrid SIFT-SVM is proposed. The current database is divided into two parts; training and testing database. The SIFT feature will be created for each training images and the keypoints are computed; then the SVM is applied for the matching process for test images. Results are obtained for three cases child; adult and old age which are made on the basis of age. The recognition rate has been computed by False Acceptance Rate (FAR) and False Rejection Rate (FRR) on these cases and then the results are compared with other algorithms. The recognized result provides robust performance under various conditions like different pose; lighting conditions and facial expressions.