The challenges of high intra-class variance yet low inter-class fluctuations in fine-grained visual categorization are more severe with few labeled samples, \textit{i.e.,} Fine-Grained categorization problems under the Few-Shot setting (FGFS). High-order features are usually developed to uncover subtle differences between sub-categories in FGFS, but they are less effective in handling the high intra-class variance. In this paper, we propose a Target-Oriented Alignment Network (TOAN) to investigate the fine-grained relation between the target query image and support classes. The feature of each support image is transformed to match the query ones in the embedding feature space, which reduces the disparity explicitly within each category. Moreover, different from existing FGFS approaches devise the high-order features over the global image with less explicit consideration of discriminative parts, we generate discriminative fine-grained features by integrating compositional concept representations to global second-order pooling. Extensive experiments are conducted on four fine-grained benchmarks to demonstrate the effectiveness of TOAN compared with the state-of-the-art models.
Detection and description of keypoints from an image is a well-studied problem in Computer Vision. Some methods like SIFT, SURF or ORB are computationally really efficient. This paper proposes a solution for a particular case study on object recognition of industrial parts based on hierarchical classification. Reducing the number of instances leads to better performance, indeed, that is what the use of the hierarchical classification is looking for. We demonstrate that this method performs better than using just one method like ORB, SIFT or FREAK, despite being fairly slower.
Biometric methods based on iris images are believed to allow very high accuracy, and there has been an explosion of interest in iris biometrics in recent years. In this paper, we use the Scale Invariant Feature Transformation (SIFT) for recognition using iris images. Contrarily to traditional iris recognition systems, the SIFT approach does not rely on the transformation of the iris pattern to polar coordinates or on highly accurate segmentation, allowing less constrained image acquisition conditions. We extract characteristic SIFT feature points in scale space and perform matching based on the texture information around the feature points using the SIFT operator. Experiments are done using the BioSec multimodal database, which includes 3,200 iris images from 200 individuals acquired in two different sessions. We contribute with the analysis of the influence of different SIFT parameters on the recognition performance. We also show the complementarity between the SIFT approach and a popular matching approach based on transformation to polar coordinates and Log-Gabor wavelets. The combination of the two approaches achieves significantly better performance than either of the individual schemes, with a performance improvement of 24% in the Equal Error Rate.
Differentially-Private Stochastic Gradient Descent (DP-SGD) prevents training-data privacy breaches by adding noise to the clipped gradient during SGD training to satisfy the differential privacy (DP) definition. On the other hand, the same clipping operation and additive noise across training steps results in unstable updates and even a ramp-up period, which significantly reduces the model's accuracy. In this paper, we extend the Gaussian DP central limit theorem to calibrate the clipping value and the noise power for each individual step separately. We, therefore, are able to propose the dynamic DP-SGD, which has a lower privacy cost than the DP-SGD during updates until they achieve the same target privacy budget at a target number of updates. Dynamic DP-SGD, in particular, improves model accuracy without sacrificing privacy by gradually lowering both clipping value and noise power while adhering to a total privacy budget constraint. Extensive experiments on a variety of deep learning tasks, including image classification, natural language processing, and federated learning, show that the proposed dynamic DP-SGD algorithm stabilizes updates and, as a result, significantly improves model accuracy in the strong privacy protection region when compared to DP-SGD.
Recently, generative adversarial networks (GAN) have gathered a lot of interest. Their efficiency in generating unseen samples of high quality, especially images, has improved over the years. In the field of Natural Language Generation (NLG), the use of the adversarial setting to generate meaningful sentences has shown to be difficult for two reasons: the lack of existing architectures to produce realistic sentences and the lack of evaluation tools. In this paper, we propose an adversarial architecture related to the conditional GAN (cGAN) that generates sentences according to a given image (also called image captioning). This attempt is the first that uses no pre-training or reinforcement methods. We also explain why our experiment settings can be safely evaluated and interpreted for further works.
Deep neural networks (DNNs) have become remarkably successful in data prediction, and have even been used to predict future actions based on limited input. This raises the question: do these systems actually "understand" the event similar to humans? Here, we address this issue using videos taken from an accident situation in a driving simulation. In this situation, drivers had to choose between crashing into a suddenly-appeared obstacle or steering their car off a previously indicated cliff. We compared how well humans and a DNN predicted this decision as a function of time before the event. The DNN outperformed humans for early time-points, but had an equal performance for later time-points. Interestingly, spatio-temporal image manipulations and Grad-CAM visualizations uncovered some expected behavior, but also highlighted potential differences in temporal processing for the DNN.
In contrast to image/text data whose order can be used to perform non-local feature aggregation in a straightforward way using the pooling layers, graphs lack the tensor representation and mostly the element-wise max/mean function is utilized to aggregate the locally extracted feature vectors. In this paper, we present a novel approach for global feature aggregation in Graph Neural Networks (GNNs) which utilizes a Latent Fixed Data Structure (LFDS) to aggregate the extracted feature vectors. The locally extracted feature vectors are sorted/distributed on the LFDS and a latent neural network (CNN/GNN) is utilized to perform feature aggregation on the LFDS. The proposed approach is used to design several novel global feature aggregation methods based on the choice of the LFDS. We introduce multiple LFDSs including loop, 3D tensor (image), sequence, data driven graphs and an algorithm which sorts/distributes the extracted local feature vectors on the LFDS. While the computational complexity of the proposed methods are linear with the order of input graphs, they achieve competitive or better results.
Adversarial patch attack aims to fool a machine learning model by arbitrarily modifying pixels within a restricted region of an input image. Such attacks are a major threat to models deployed in the physical world, as they can be easily realized by presenting a customized object in the camera view. Defending against such attacks is challenging due to the arbitrariness of patches, and existing provable defenses suffer from poor certified accuracy. In this paper, we propose PatchVeto, a zero-shot certified defense against adversarial patches based on Vision Transformer (ViT) models. Rather than training a robust model to resist adversarial patches which may inevitably sacrifice accuracy, PatchVeto reuses a pretrained ViT model without any additional training, which can achieve high accuracy on clean inputs while detecting adversarial patched inputs by simply manipulating the attention map of ViT. Specifically, each input is tested by voting over multiple inferences with different attention masks, where at least one inference is guaranteed to exclude the adversarial patch. The prediction is certifiably robust if all masked inferences reach consensus, which ensures that any adversarial patch would be detected with no false negative. Extensive experiments have shown that PatchVeto is able to achieve high certified accuracy (e.g. 67.1% on ImageNet for 2%-pixel adversarial patches), significantly outperforming state-of-the-art methods. The clean accuracy is the same as vanilla ViT models (81.8% on ImageNet) since the model parameters are directly reused. Meanwhile, our method can flexibly handle different adversarial patch sizes by simply changing the masking strategy.
In recent years, convolutional neural networks (CNNs) have revolutionized medical image analysis. One of the most well-known CNN architectures in semantic segmentation is the U-net, which has achieved much success in several medical image segmentation applications. Also more recently, with the rise of autoML ad advancements in neural architecture search (NAS), methods like NAS-Unet have been proposed for NAS in medical image segmentation. In this paper, with inspiration from LadderNet, U-Net, autoML and NAS, we propose an ensemble deep neural network with an underlying U-Net framework consisting of bi-directional convolutional LSTMs and dense connections, where the first (from left) U-Net-like network is deeper than the second (from left). We show that this ensemble network outperforms recent state-of-the-art networks in several evaluation metrics, and also evaluate a lightweight version of this ensemble network, which also outperforms recent state-of-the-art networks in some evaluation metrics.
Deep learning techniques have led to state-of-the-art single image super-resolution (SISR) with natural images. Pairs of high-resolution (HR) and low-resolution (LR) images are used to train the deep learning model (mapping function). These techniques have also been applied to medical image super-resolution (SR). Compared with natural images, medical images have several unique characteristics. First, there are no HR images for training in real clinical applications because of the limitations of imaging systems and clinical requirements. Second, other modal HR images are available (e.g., HR T1-weighted images are available for enhancing LR T2-weighted images). In this paper, we propose an unsupervised SISR technique based on simple prior knowledge of the human anatomy; this technique does not require HR images for training. Furthermore, we present a guided residual dense network, which incorporates a residual dense network with a guided deep convolutional neural network for enhancing the resolution of LR images by referring to different HR images of the same subject. Experiments on a publicly available brain MRI database showed that our proposed method achieves better performance than the state-of-the-art methods.