arXiv subjects
Publications and source records attributed to Yue Wu.
A randomised trapezoidal quadrature rule is proposed for continuous functions which enjoys less regularity than commonly required. Indeed, we consider functions in some fractional Sobolev space. Various error bounds for this randomised rule are established while an error bound for classical trapezoidal quadrature is obtained for comparison. The randomised trapezoidal quadrature rule is shown to improve the order of convergence by half.
Object detection models perform well at localizing and classifying objects that they are shown during training. However, due to the difficulty and cost associated with creating and annotating detection datasets, trained models detect a limited number of object types with unknown objects treated as background content. This hinders the adoption of conventional detectors in real-world applications like large-scale object matching, visual grounding, visual relation prediction, obstacle detection (where it is more important to determine the presence and location of objects than to find specific types), etc. We propose class-agnostic object detection as a new problem that focuses on detecting objects irrespective of their object-classes. Specifically, the goal is to predict bounding boxes for all objects in an image but not their object-classes. The predicted boxes can then be consumed by another system to perform application-specific classification, retrieval, etc. We propose training and evaluation protocols for benchmarking class-agnostic detectors to advance future research in this domain. Finally, we propose (1) baseline methods and (2) a new adversarial learning framework for class-agnostic detection that forces the model to exclude class-specific information from features used for predictions. Experimental results show that adversarial learning improves class-agnostic detection efficacy.
Advances in visual navigation methods have led to intelligent embodied navigation agents capable of learning meaningful representations from raw RGB images and perform a wide variety of tasks involving structural and semantic reasoning. However, most learning-based navigation policies are trained and tested in simulation environments. In order for these policies to be practically useful, they need to be transferred to the real-world. In this paper, we propose an unsupervised domain adaptation method for visual navigation. Our method translates the images in the target domain to the source domain such that the translation is consistent with the representations learned by the navigation policy. The proposed method outperforms several baselines across two different navigation tasks in simulation. We further show that our method can be used to transfer the navigation policies learned in simulation to the real world.
Despite success on a wide range of problems related to vision, generative adversarial networks (GANs) often suffer from inferior performance due to unstable training, especially for text generation. To solve this issue, we propose a new variational GAN training framework which enjoys superior training stability. Our approach is inspired by a connection of GANs and reinforcement learning under a variational perspective. The connection leads to (1) probability ratio clipping that regularizes generator training to prevent excessively large updates, and (2) a sample re-weighting mechanism that improves discriminator training by downplaying bad-quality fake samples. Moreover, our variational GAN framework can provably overcome the training issue in many GANs that an optimal discriminator cannot provide any informative gradient to training generator. By plugging the training approach in diverse state-of-the-art GAN architectures, we obtain significantly improved performance over a range of tasks, including text generation, text style transfer, and image generation.
Sturge-Weber syndrome (SWS) is a vascular malformation disease, and it may cause blindness if the patient's condition is severe. Clinical results show that SWS can be divided into two types based on the characteristics of scleral blood vessels. Therefore, how to accurately segment scleral blood vessels has become a significant problem in computer-aided diagnosis. In this research, we propose to continuously upsample the bottom layer's feature maps to preserve image details, and design a novel Claw UNet based on UNet for scleral blood vessel segmentation. Specifically, the residual structure is used to increase the number of network layers in the feature extraction stage to learn deeper features. In the decoding stage, by fusing the features of the encoding, upsampling, and decoding parts, Claw UNet can achieve effective segmentation in the fine-grained regions of scleral blood vessels. To effectively extract small blood vessels, we use the attention mechanism to calculate the attention coefficient of each position in images. Claw UNet outperforms other UNet-based networks on scleral blood vessel image dataset.
In this paper we put the visibility transformation on a clear theoretical footing and show that this transform is able to embed the effect of the absolute position of the data stream into signature features in a unified and efficient way. The generated feature set is particularly useful in pattern recognition tasks, for its simplifying role in allowing the signature feature set to accommodate nonlinear functions of absolute and relative values.
An intriguing phenomenon observed during training neural networks is the spectral bias, which states that neural networks are biased towards learning less complex functions. The priority of learning functions with low complexity might be at the core of explaining generalization ability of neural network, and certain efforts have been made to provide theoretical explanation for spectral bias. However, there is still no satisfying theoretical result justifying the underlying mechanism of spectral bias. In this paper, we give a comprehensive and rigorous explanation for spectral bias and relate it with the neural tangent kernel function proposed in recent work. We prove that the training process of neural networks can be decomposed along different directions defined by the eigenfunctions of the neural tangent kernel, where each direction has its own convergence rate and the rate is determined by the corresponding eigenvalue. We then provide a case study when the input data is uniformly distributed over the unit sphere, and show that lower degree spherical harmonics are easier to be learned by over-parameterized neural networks. Finally, we provide numerical experiments to demonstrate the correctness of our theory. Our experimental results also show that our theory can tolerate certain model misspecification in terms of the input data distribution.
Planimation is a modular and extensible open source framework to visualise sequential solutions of planning problems specified in PDDL. We introduce a preliminary declarative PDDL-like animation profile specification, expressive enough to synthesise animations of arbitrary initial states and goals of a benchmark with just a single profile.
The Teacher-Student (T-S) framework is widely utilized in the classification tasks, through which the performance of one neural network (the student) can be improved by transferring knowledge from another trained neural network (the teacher). Since the transferring knowledge is related to the network capacities and structures between the teacher and the student, how to define efficient knowledge remains an open question. To address this issue, we design a novel transferring knowledge, the Self-Attention based Inter-Class Correlation (ICC) map in the output layer, and propose our T-S framework, Inter-Class Correlation Transfer (ICCT).
Detecting anomalous edges and nodes in dynamic networks is critical in various areas, such as social media, computer networks, and so on. Recent approaches leverage network embedding technique to learn how to generate node representations for normal training samples and detect anomalies deviated from normal patterns. However, most existing network embedding approaches learn deterministic node representations, which are sensitive to fluctuations of the topology and attributes due to the high flexibility and stochasticity of dynamic networks. In this paper, a stochastic neural network, named by Hierarchical Variational Graph Recurrent Autoencoder (H-VGRAE), is proposed to detect anomalies in dynamic networks by the learned robust node representations in the form of random variables. H-VGRAE is a semi-supervised model to capture normal patterns in training set by maximizing the likelihood of the adjacency matrix and node attributes via variational inference. Comparing with existing methods, H-VGRAE has three main advantages: 1) H-VGRAE learns robust node representations through stochasticity modeling and the extraction of multi-scale spatial-temporal features; 2) H-VGRAE can be extended to deep structure with the increase of the dynamic network scale; 3) the anomalous edge and node can be located and interpreted from the probabilistic perspective. Extensive experiments on four real-world datasets demonstrate the outperformance of H-VGRAE on anomaly detection in dynamic networks compared with state-of-the-art competitors.
The availability of mobile technologies has enabled the efficient collection prospective longitudinal, ecologically valid self-reported mood data from psychiatric patients. These data streams have potential for improving the efficiency and accuracy of psychiatric diagnosis as well predicting future mood states enabling earlier intervention. However, missing responses are common in such datasets and there is little consensus as to how this should be dealt with in practice. A signature-based method was used to capture different elements of self-reported mood alongside missing data to both classify diagnostic group and predict future mood in patients with bipolar disorder, borderline personality disorder and healthy controls. The missing-response-incorporated signature-based method achieves roughly 66\% correct diagnosis, with f1 scores for three different clinic groups 59\% (bipolar disorder), 75\% (healthy control) and 61\% (borderline personality disorder) respectively. This was significantly more efficient than the naive model which excluded missing data. Accuracies of predicting subsequent mood states and scores were also improved by inclusion of missing responses. The signature method provided an effective approach to the analysis of prospectively collected mood data where missing data was common and should be considered as an approach in other similar datasets.
This paper is devoted to the $L^p(\mathbb R)$ theory of the fractional Fourier transform (FRFT) for $1\le p < 2$. In view of the special structure of the FRFT, we study FRFT properties of $L^1$ functions, via the introduction of a suitable chirp operator. However, in the $L^1(\mathbb{R})$ setting, problems of convergence arise even when basic manipulations of functions are performed. We overcome such issues and study the FRFT inversion problem via approximation by suitable means, such as the fractional Gauss and Abel means. We also obtain the regularity of fractional convolution and results on pointwise convergence of FRFT means. Finally we discuss $L^p$ multiplier results and a Littlewood-Paley theorem associated with FRFT.
We study a class of semi-implicit Taylor-type numerical methods that are easy to implement and designed to solve multidimensional stochastic differential equations driven by a general rough noise, e.g. a fractional Brownian motion. In the multiplicative noise case, the equation is understood as a rough differential equation in the sense of T.~Lyons. We focus on equations for which the drift coefficient may be unbounded and satisfies a one-sided Lipschitz condition only. We prove well-posedness of the methods, provide a full analysis, and deduce their convergence rate. Numerical experiments show that our schemes are particularly useful in the case of stiff rough stochastic differential equations driven by a fractional Brownian motion.
Theoretical stellar spectra rely on model stellar atmospheres computed based on our understanding of the physical laws at play in the stellar interiors. These models, coupled with atomic and molecular line databases, are used to generate theoretical stellar spectral libraries (SSLs) comprising of stellar spectra over a regular grid of atmospheric parameters (temperature, surface gravity, abundances) at any desired resolution. Another class of SSLs is referred to as empirical spectral libraries; these contain observed spectra at limited resolution. SSLs play an essential role in deriving the properties of stars and stellar populations. Both theoretical and empirical libraries suffer from limited coverage over the parameter space. This limitation is overcome to some extent by generating spectra for specific sets of atmospheric parameters by interpolating within the grid of available parameter space. In this work, we present a method for spectral interpolation in the optical region using machine learning algorithms that are generic, easily adaptable for any SSL without much change in the model parameters, and computationally inexpensive. We use two machine learning techniques, Random Forest (RF) and Artificial Neural Networks (ANN), and train the models on the MILES library. We apply the trained models to spectra from the CFLIB for testing and show that the performance of the two models is comparable. We show that both the models achieve better accuracy than the existing methods of polynomial based interpolation and the Gaussian radial basis function (RBF) interpolation.
Inspired by the latest developments in multilevel Monte Carlo (MLMC) methods and randomised sketching for linear algebra problems we propose a MLMC estimator for real-time processing of matrix structured random data. Our algorithm is particularly effective in handling high-dimensional inner products and matrix multiplication, in applications of image analysis and large-scale supervised learning.
We present an approach to predict future video frames given a sequence of continuous video frames in the past. Instead of synthesizing images directly, our approach is designed to understand the complex scene dynamics by decoupling the background scene and moving objects. The appearance of the scene components in the future is predicted by non-rigid deformation of the background and affine transformation of moving objects. The anticipated appearances are combined to create a reasonable video in the future. With this procedure, our method exhibits much less tearing or distortion artifact compared to other approaches. Experimental results on the Cityscapes and KITTI datasets show that our model outperforms the state-of-the-art in terms of visual quality and accuracy.
Two head structures (i.e. fully connected head and convolution head) have been widely used in R-CNN based detectors for classification and localization tasks. However, there is a lack of understanding of how does these two head structures work for these two tasks. To address this issue, we perform a thorough analysis and find an interesting fact that the two head structures have opposite preferences towards the two tasks. Specifically, the fully connected head (fc-head) is more suitable for the classification task, while the convolution head (conv-head) is more suitable for the localization task. Furthermore, we examine the output feature maps of both heads and find that fc-head has more spatial sensitivity than conv-head. Thus, fc-head has more capability to distinguish a complete object from part of an object, but is not robust to regress the whole object. Based upon these findings, we propose a Double-Head method, which has a fully connected head focusing on classification and a convolution head for bounding box regression. Without bells and whistles, our method gains +3.5 and +2.8 AP on MS COCO dataset from Feature Pyramid Network (FPN) baselines with ResNet-50 and ResNet-101 backbones, respectively.
Cross-modality person re-identification (cm-ReID) is a challenging but key technology for intelligent video analysis. Existing works mainly focus on learning common representation by embedding different modalities into a same feature space. However, only learning the common characteristics means great information loss, lowering the upper bound of feature distinctiveness. In this paper, we tackle the above limitation by proposing a novel cross-modality shared-specific feature transfer algorithm (termed cm-SSFT) to explore the potential of both the modality-shared information and the modality-specific characteristics to boost the re-identification performance. We model the affinities of different modality samples according to the shared features and then transfer both shared and specific features among and across modalities. We also propose a complementary feature learning strategy including modality adaption, project adversarial learning and reconstruction enhancement to learn discriminative and complementary shared and specific features of each modality, respectively. The entire cm-SSFT algorithm can be trained in an end-to-end manner. We conducted comprehensive experiments to validate the superiority of the overall algorithm and the effectiveness of each component. The proposed algorithm significantly outperforms state-of-the-arts by 22.5% and 19.3% mAP on the two mainstream benchmark datasets SYSU-MM01 and RegDB, respectively.