Multivariate time series analysis is an important problem in data mining because of its widespread applications. With the increase of time series data available for training, implementing deep neural networks in the field of time series analysis is becoming common. Res2Net, a recently proposed backbone, can further improve the state-of-the-art networks as it improves the multi-scale representation ability through connecting different groups of filters. However, Res2Net ignores the correlations of the feature maps and lacks the control on the information interaction process. To address that problem, in this paper, we propose a backbone convolutional neural network based on the thought of gated mechanism and Res2Net, namely Gated Res2Net (GRes2Net), for multivariate time series analysis. The hierarchical residual-like connections are influenced by gates whose values are calculated based on the original feature maps, the previous output feature maps and the next input feature maps thus considering the correlations between the feature maps more effectively. Through the utilization of gated mechanism, the network can control the process of information sending hence can better capture and utilize the both the temporal information and the correlations between the feature maps. We evaluate the GRes2Net on four multivariate time series datasets including two classification datasets and two forecasting datasets. The results demonstrate that GRes2Net have better performances over the state-of-the-art methods thus indicating the superiority
Deep encoder-decoder based CNNs have advanced image inpainting methods for hole filling. While existing methods recover structures and textures step-by-step in the hole regions, they typically use two encoder-decoders for separate recovery. The CNN features of each encoder are learned to capture either missing structures or textures without considering them as a whole. The insufficient utilization of these encoder features limit the performance of recovering both structures and textures. In this paper, we propose a mutual encoder-decoder CNN for joint recovery of both. We use CNN features from the deep and shallow layers of the encoder to represent structures and textures of an input image, respectively. The deep layer features are sent to a structure branch and the shallow layer features are sent to a texture branch. In each branch, we fill holes in multiple scales of the CNN features. The filled CNN features from both branches are concatenated and then equalized. During feature equalization, we reweigh channel attentions first and propose a bilateral propagation activation function to enable spatial equalization. To this end, the filled CNN features of structure and texture mutually benefit each other to represent image content at all feature levels. We use the equalized feature to supplement decoder features for output image generation through skip connections. Experiments on the benchmark datasets show the proposed method is effective to recover structures and textures and performs favorably against state-of-the-art approaches.
This paper targets on learning-based novel view synthesis from a single or limited 2D images without the pose supervision. In the viewer-centered coordinates, we construct an end-to-end trainable conditional variational framework to disentangle the unsupervisely learned relative-pose/rotation and implicit global 3D representation (shape, texture and the origin of viewer-centered coordinates, etc.). The global appearance of the 3D object is given by several appearance-describing images taken from any number of viewpoints. Our spatial correlation module extracts a global 3D representation from the appearance-describing images in a permutation invariant manner. Our system can achieve implicitly 3D understanding without explicitly 3D reconstruction. With an unsupervisely learned viewer-centered relative-pose/rotation code, the decoder can hallucinate the novel view continuously by sampling the relative-pose in a prior distribution. In various applications, we demonstrate that our model can achieve comparable or even better results than pose/3D model-supervised learning-based novel view synthesis (NVS) methods with any number of input views.
Co-registering the Sentinel-1 SAR and Sentinel-2 optical data of European Space Agency (ESA) is of great importance for many remote sensing applications. The Sentinel-1 and 2 product specifications from ESA show that the Sentinel-1 SAR L1 and the Sentinel-2 optical L1C images have a co-registration accuracy of within 2 pixels. However, we find that the actual misregistration errors are much larger than that between such images. This paper measures the misregistration errors by a block-based multimodal image matching strategy to six pairs of the Sentinel-1 SAR and Sentinel-2 optical images, which locate in China and Europe and cover three different terrains such as flat areas, hilly areas and mountainous areas. Our experimental results show the misregistration errors of the flat areas are 20-30 pixels, and these of the hilly areas are 20-40 pixels. While in the mountainous areas, the errors increase to 50-60 pixels. To eliminate the misregistration, we use some representative geometric transformation models such as polynomial models, projective models, and rational function models for the co-registration of the two types of images, and compare and analyze their registration accuracy under different number of control points and different terrains. The results of our analysis show that the 3rd. Order polynomial achieves the most satisfactory registration results. Its registration accuracy of the flat areas is less than 1.0 10m pixels, and that of the hilly areas is about 1.5 pixels, and that of the mountainous areas is between 1.8 and 2.3 pixels. In a word, this paper discloses and measures the misregistration between the Sentinel-1 SAR L1 and Sentinel-2 optical L1C images for the first time. Moreover, we also determine a relatively optimal geometric transformation model of the co-registration of the two types of images.
Co-Registration of aerial imagery and Light Detection and Ranging (LiDAR) data is quilt challenging because the different imaging mechanism causes significant geometric and radiometric distortions between such data. To tackle the problem, this paper proposes an automatic registration method based on structural features and three-dimension (3D) phase correlation. In the proposed method, the LiDAR point cloud data is first transformed into the intensity map, which is used as the reference image. Then, we employ the Fast operator to extract uniformly distributed interest points in the aerial image by a partition strategy and perform a local geometric correction by using the collinearity equation to eliminate scale and rotation difference between images. Subsequently, a robust structural feature descriptor is build based on dense gradient features, and the 3D phase correlation is used to detect control points (CPs) between aerial images and LiDAR data in the frequency domain, where the image matching is accelerated by the 3D Fast Fourier Transform (FFT). Finally, the obtained CPs are employed to correct the exterior orientation elements, which is used to achieve co-registration of aerial images and LiDAR data. Experiments with two datasets of aerial images and LiDAR data show that the proposed method is much faster and more robust than state of the art methods
The truncated Tucker decomposition, also known as the truncated higher-order singular value decomposition (HOSVD), has been extensively utilized as an efficient tool in many applications. Popular direct methods for truncated HOSVD often suffer from the notorious intermediate data explosion issue and are not easy to parallelize. In this paper, we propose a class of new truncated HOSVD algorithms based on alternating least squares (ALS). The proposed ALS-based approaches are able to eliminate the redundant computations of the singular vectors of intermediate matrices and are therefore free of data explosion. Also, the new methods are more flexible with adjustable convergence tolerance and are intrinsically parallelizable on high-performance computers. Theoretical analysis reveals that the ALS iteration in the proposed algorithms is q-linear convergent with a relatively wide convergence region. Numerical experiments with both synthetic and real-world tensor data demonstrate that ALS-based methods can substantially reduce the total cost of the original ones and are highly scalable for parallel computing.
We present PFNN, a penalty-free neural network method, to efficiently solve a class of second-order boundary-value problems on complex geometries. To reduce the smoothness requirement, the original problem is reformulated to a weak form so that the evaluations of high-order derivatives are avoided. Two neural networks, rather than just one, are employed to construct the approximate solution, with one network satisfying the essential boundary conditions and the other handling the rest part of the domain. In this way, an unconstrained optimization problem, instead of a constrained one, is solved without adding any penalty terms. The entanglement of the two networks is eliminated with the help of a length factor function that is scale invariant and can adapt with complex geometries. We prove the convergence of the PFNN method and conduct numerical experiments on a series of linear and nonlinear second-order boundary-value problems to demonstrate that PFNN is superior to several existing approaches in terms of accuracy, flexibility and robustness.
In this paper, we propose a framework capable of generating face images that fall into the same distribution as that of a given one-shot example. We leverage a pre-trained StyleGAN model that already learned the generic face distribution. Given the one-shot target, we develop an iterative optimization scheme that rapidly adapts the weights of the model to shift the output's high-level distribution to the target's. To generate images of the same distribution, we introduce a style-mixing technique that transfers the low-level statistics from the target to faces randomly generated with the model. With that, we are able to generate an unlimited number of faces that inherit from the distribution of both generic human faces and the one-shot example. The newly generated faces can serve as augmented training data for other downstream tasks. Such setting is appealing as it requires labeling very few, or even one example, in the target domain, which is often the case of real-world face manipulations that result from a variety of unknown and unique distributions, each with extremely low prevalence. We show the effectiveness of our one-shot approach for detecting face manipulations and compare it with other few-shot domain adaptation methods qualitatively and quantitatively.
Recent deep learning based image inpainting methods which utilize contextual information and two-stage architecture have exhibited remarkable performance. However, the two-stage architecture is time-consuming, the contextual information lack high-level semantics and ignores both the semantic relevance and distance information of hole's feature patches, these limitations result in blurry textures and distorted structures of final result. Motivated by these observations, we propose a new deep generative model-based approach, which trains a shared network twice with different targets and utilizes a single network during the testing phase, so that we can effectively save inference time. Specifically, the targets of two training steps are structure reconstruction and texture generation respectively. During the second training, we first propose a Pyramid Filling Block (PF-block) to utilize the high-level features that the hole regions has been filled to guide the filling process of low-level features progressively, the missing content can be filled from deep to shallow in a pyramid fashion. Then, inspired by the classical bilateral filter [30], we propose the Bilateral Attention layer (BA-layer) to optimize filled feature map, which synthesizes feature patches at each position by computing weighted sums of the surrounding feature patches, these weights are derived by considering both distance and value relationships between feature patches, thus making the visually plausible inpainting results. Finally, experiments on multiple publicly available datasets show the superior performance of our approach.