arXiv subjects

Ping Wang

Publications and source records attributed to Ping Wang.

At least 235 records · Page 13Linked to original sources

arXiv2021

Context-aware virtual adversarial training for anatomically-plausible segmentation

Despite their outstanding accuracy, semi-supervised segmentation methods based on deep neural networks can still yield predictions that are considered anatomically impossible by clinicians, for instance, containing holes or disconnected regions. To solve this problem, we present a Context-aware Virtual Adversarial Training (CaVAT) method for generating anatomically plausible segmentation. Unlike approaches focusing solely on accuracy, our method also considers complex topological constraints like connectivity which cannot be easily modeled in a differentiable loss function. We use adversarial training to generate examples violating the constraints, so the network can learn to avoid making such incorrect predictions on new examples, and employ the Reinforce algorithm to handle non-differentiable segmentation constraints. The proposed method offers a generic and efficient way to add any constraint on top of any segmentation network. Experiments on two clinically-relevant datasets show our method to produce segmentations that are both accurate and anatomically-plausible in terms of region connectivity.

cs.CV↗

arXiv2021

Enhancing Geometric Factors in Model Learning and Inference for Object Detection and Instance Segmentation

Deep learning-based object detection and instance segmentation have achieved unprecedented progress. In this paper, we propose Complete-IoU (CIoU) loss and Cluster-NMS for enhancing geometric factors in both bounding box regression and Non-Maximum Suppression (NMS), leading to notable gains of average precision (AP) and average recall (AR), without the sacrifice of inference efficiency. In particular, we consider three geometric factors, i.e., overlap area, normalized central point distance and aspect ratio, which are crucial for measuring bounding box regression in object detection and instance segmentation. The three geometric factors are then incorporated into CIoU loss for better distinguishing difficult regression cases. The training of deep models using CIoU loss results in consistent AP and AR improvements in comparison to widely adopted $\ell_n$-norm loss and IoU-based loss. Furthermore, we propose Cluster-NMS, where NMS during inference is done by implicitly clustering detected boxes and usually requires less iterations. Cluster-NMS is very efficient due to its pure GPU implementation, and geometric factors can be incorporated to improve both AP and AR. In the experiments, CIoU loss and Cluster-NMS have been applied to state-of-the-art instance segmentation (e.g., YOLACT and BlendMask-RT), and object detection (e.g., YOLO v3, SSD and Faster R-CNN) models. Taking YOLACT on MS COCO as an example, our method achieves performance gains as +1.7 AP and +6.2 AR$_{100}$ for object detection, and +0.9 AP and +3.5 AR$_{100}$ for instance segmentation, with 27.1 FPS on one NVIDIA GTX 1080Ti GPU. All the source code and trained models are available at https://github.com/Zzh-tju/CIoU

cs.CV↗

arXiv2021

Corpus-level and Concept-based Explanations for Interpretable Document Classification

Using attention weights to identify information that is important for models' decision-making is a popular approach to interpret attention-based neural networks. This is commonly realized in practice through the generation of a heat-map for every single document based on attention weights. However, this interpretation method is fragile, and easy to find contradictory examples. In this paper, we propose a corpus-level explanation approach, which aims to capture causal relationships between keywords and model predictions via learning the importance of keywords for predicted labels across a training corpus based on attention weights. Based on this idea, we further propose a concept-based explanation method that can automatically learn higher-level concepts and their importance to model prediction tasks. Our concept-based explanation method is built upon a novel Abstraction-Aggregation Network, which can automatically cluster important keywords during an end-to-end training process. We apply these methods to the document classification task and show that they are powerful in extracting semantically meaningful keywords and concepts. Our consistency analysis results based on an attention-based Naïve Bayes classifier also demonstrate these keywords and concepts are important for model predictions.

cs.IR↗

arXiv2021

An Interpretable and Uncertainty Aware Multi-Task Framework for Multi-Aspect Sentiment Analysis

In recent years, several online platforms have seen a rapid increase in the number of review systems that request users to provide aspect-level feedback. Document-level Multi-aspect Sentiment Classification (DMSC), where the goal is to predict the ratings/sentiment from a review at an individual aspect level, has become a challenging and imminent problem. To tackle this challenge, we propose a deliberate self-attention-based deep neural network model, namely FEDAR, for the DMSC problem, which can achieve competitive performance while also being able to interpret the predictions made. FEDAR is equipped with a highway word embedding layer to transfer knowledge from pre-trained word embeddings, an RNN encoder layer with output features enriched by pooling and factorization techniques, and a deliberate self-attention layer. In addition, we also propose an Attention-driven Keywords Ranking (AKR) method, which can automatically discover aspect keywords and aspect-level opinion keywords from the review corpus based on the attention weights. These keywords are significant for rating predictions by FEDAR. Since crowdsourcing annotation can be an alternate way to recover missing ratings of reviews, we propose a LEcture-AuDience (LEAD) strategy to estimate model uncertainty in the context of multi-task learning, so that valuable human resources can focus on the most uncertain predictions. Our extensive set of experiments on five different open-domain DMSC datasets demonstrate the superiority of the proposed FEDAR and LEAD models. We further introduce two new DMSC datasets in the healthcare domain and benchmark different baseline models and our models on them. Attention weights visualization results and visualization of aspect and opinion keywords demonstrate the interpretability of our model and the effectiveness of our AKR method.

cs.CL↗

arXiv2021

CNN Retrieval based Unsupervised Metric Learning for Near-Duplicated Video Retrieval

As important data carriers, the drastically increasing number of multimedia videos often brings many duplicate and near-duplicate videos in the top results of search. Near-duplicate video retrieval (NDVR) can cluster and filter out the redundant contents. In this paper, the proposed NDVR approach extracts the frame-level video representation based on convolutional neural network (CNN) features from fully-connected layer and aggregated intermediate convolutional layers. Unsupervised metric learning is used for similarity measurement and feature matching. An efficient re-ranking algorithm combined with k-nearest neighborhood fuses the retrieval results from two levels of features and further improves the retrieval performance. Extensive experiments on the widely used CC\_WEB\_VIDEO dataset shows that the proposed approach exhibits superior performance over the state-of-the-art.

cs.IR↗

arXiv2021

Restoring the top-of-atmosphere reflectance during solar eclipses: a proof of concept with the UV Absorbing Aerosol Index measured by TROPOMI

During a solar eclipse the solar irradiance reaching the top-of-atmosphere (TOA) is reduced in the Moon shadow. The solar irradiance is commonly measured by Earth observation satellites before the start of the solar eclipse and is not corrected for this reduction, which results in a decrease of the computed TOA reflectances. Consequently, air quality products that are derived from TOA reflectance spectra, such as the ultraviolet (UV) Absorbing Aerosol Index (AAI), are distorted or undefined in the shadow of the Moon. The availability of air quality satellite data in the penumbral and antumbral shadow during solar eclipses, however, is of particular interest to users studying the atmospheric response to solar eclipses. Given the time and location of a point on the Earth's surface, we explain how to compute the obscuration during a solar eclipse taking into account wavelength-dependent solar limb darkening. With the calculated obscuration fractions, we restore the TOA reflectances and the AAI in the penumbral shadow during the annular solar eclipses on 26 December 2019 and 21 June 2020 measured by the TROPOMI/S5P instrument. In the corrected products, the signature of the Moon shadow disappeared, but only if wavelength-dependent solar limb darkening is taken into account. We conclude that the correction method of this paper can be used to detect real AAI rising phenomena during a solar eclipse and has the potential to restore any other product that is derived from TOA reflectance spectra. This would resolve the solar eclipse anomalies in satellite air quality measurements and would allow for studying the effect of the eclipse obscuration on the composition of the Earth's atmosphere from space.

physics.ao-ph↗

arXiv2021

Aligning Visual Prototypes with BERT Embeddings for Few-Shot Learning

Few-shot learning (FSL) is the task of learning to recognize previously unseen categories of images from a small number of training examples. This is a challenging task, as the available examples may not be enough to unambiguously determine which visual features are most characteristic of the considered categories. To alleviate this issue, we propose a method that additionally takes into account the names of the image classes. While the use of class names has already been explored in previous work, our approach differs in two key aspects. First, while previous work has aimed to directly predict visual prototypes from word embeddings, we found that better results can be obtained by treating visual and text-based prototypes separately. Second, we propose a simple strategy for learning class name embeddings using the BERT language model, which we found to substantially outperform the GloVe vectors that were used in previous work. We furthermore propose a strategy for dealing with the high dimensionality of these vectors, inspired by models for aligning cross-lingual word embeddings. We provide experiments on miniImageNet, CUB and tieredImageNet, showing that our approach consistently improves the state-of-the-art in metric-based FSL.

cs.CV↗

arXiv2021

Towards Strengthening Deep Learning-based Side Channel Attacks with Mixup

In recent years, various deep learning techniques have been exploited in side channel attacks, with the anticipation of obtaining more appreciable attack results. Most of them concentrate on improving network architectures or putting forward novel algorithms, assuming that there are adequate profiling traces available to train an appropriate neural network. However, in practical scenarios, profiling traces are probably insufficient, which makes the network learn deficiently and compromises attack performance. In this paper, we investigate a kind of data augmentation technique, called mixup, and first propose to exploit it in deep-learning based side channel attacks, for the purpose of expanding the profiling set and facilitating the chances of mounting a successful attack. We perform Correlation Power Analysis for generated traces and original traces, and discover that there exists consistency between them regarding leakage information. Our experiments show that mixup is truly capable of enhancing attack performance especially for insufficient profiling traces. Specifically, when the size of the training set is decreased to 30% of the original set, mixup can significantly reduce acquired attacking traces. We test three mixup parameter values and conclude that generally all of them can bring about improvements. Besides, we compare three leakage models and unexpectedly find that least significant bit model, which is less frequently used in previous works, actually surpasses prevalent identity model and hamming weight model in terms of attack results.

cs.CR↗

arXiv2021

An Improved Discriminative Optimization for 3D Rigid Point Cloud Registration

The Discriminative Optimization (DO) algorithm has been proved much successful in 3D point cloud registration. In the original DO, the feature (descriptor) of two point cloud was defined as a histogram, and the element of histogram indicates the weights of scene points in "front" or "back" side of a model point. In this paper, we extended the histogram which indicate the sides from "front-back" to "front-back", "up-down", and "clockwise-anticlockwise". In addition, we reweighted the extended histogram according to the model points' distribution. We evaluated the proposed Improved DO on the Stanford Bunny and Oxford SensatUrban dataset, and compared it with six classical State-Of-The-Art point cloud registration algorithms. The experimental result demonstrates our algorithm achieves comparable performance in point registration accuracy and root-mean-sqart-error.

cs.CV↗

arXiv2021

Fast-Convergent Federated Learning with Adaptive Weighting

Federated learning (FL) enables resource-constrained edge nodes to collaboratively learn a global model under the orchestration of a central server while keeping privacy-sensitive data locally. The non-independent-and-identically-distributed (non-IID) data samples across participating nodes slow model training and impose additional communication rounds for FL to converge. In this paper, we propose Federated Adaptive Weighting (FedAdp) algorithm that aims to accelerate model convergence under the presence of nodes with non-IID dataset. We observe the implicit connection between the node contribution to the global model aggregation and data distribution on the local node through theoretical and empirical analysis. We then propose to assign different weights for updating the global model based on node contribution adaptively through each training round. The contribution of participating nodes is first measured by the angle between the local gradient vector and the global gradient vector, and then, weight is quantified by a designed non-linear mapping function subsequently. The simple yet effective strategy can reinforce positive (suppress negative) node contribution dynamically, resulting in communication round reduction drastically. Its superiority over the commonly adopted Federated Averaging (FedAvg) is verified both theoretically and experimentally. With extensive experiments performed in Pytorch and PySyft, we show that FL training with FedAdp can reduce the number of communication rounds by up to 54.1% on MNIST dataset and up to 45.4% on FashionMNIST dataset, as compared to FedAvg algorithm.

cs.LG↗

arXiv2021

Privacy-Assured Outsourcing of Compressed Sensing Reconstruction Service in Cloud

Compressed sensing (CS), breaking the constriction of Shannon-Nyquist sampling theorem, is a very promising data acquisition technique in the era of multimedia big data. However, the high complexity of CS reconstruction algorithm is a big trouble for endusers who are hardly provided with great computing power. The combination of CS and cloud has the potential of freeing endusers from the resource constraint by cleverly transforming computational workload from the local cilent to the cloud platform. As a result, the low-complexity encoding virtue of CS is fully leveraged in the resource-constrained sensing devices but its highcomplexity decoding problem is effectively addressed in cloud. It seems to be perfect but privacy and security concerns are ignored. In this paper, a secure outsourcing scheme for CS reconstruction service is proposed. Experimental results and security analyses demonstrate that the proposed scheme can restrict malicious access, verify the integrity of the recovered data, and resist brute-force attack, ciphertext-only attack, and plaintext attack.

cs.CR↗

arXiv2021

Self-Supervised Learning of Contextual Embeddings for Link Prediction in Heterogeneous Networks

Representation learning methods for heterogeneous networks produce a low-dimensional vector embedding for each node that is typically fixed for all tasks involving the node. Many of the existing methods focus on obtaining a static vector representation for a node in a way that is agnostic to the downstream application where it is being used. In practice, however, downstream tasks such as link prediction require specific contextual information that can be extracted from the subgraphs related to the nodes provided as input to the task. To tackle this challenge, we develop SLiCE, a framework bridging static representation learning methods using global information from the entire graph with localized attention driven mechanisms to learn contextual node representations. We first pre-train our model in a self-supervised manner by introducing higher-order semantic associations and masking nodes, and then fine-tune our model for a specific link prediction task. Instead of training node representations by aggregating information from all semantic neighbors connected via metapaths, we automatically learn the composition of different metapaths that characterize the context for a specific task without the need for any pre-defined metapaths. SLiCE significantly outperforms both static and contextual embedding learning methods on several publicly available benchmark network datasets. We also interpret the semantic association matrix and provide its utility and relevance in making successful link predictions between heterogeneous nodes in the network.

cs.LG↗

arXiv2021

RAGA: Relation-aware Graph Attention Networks for Global Entity Alignment

Entity alignment (EA) is the task to discover entities referring to the same real-world object from different knowledge graphs (KGs), which is the most crucial step in integrating multi-source KGs. The majority of the existing embeddings-based entity alignment methods embed entities and relations into a vector space based on relation triples of KGs for local alignment. As these methods insufficiently consider the multiple relations between entities, the structure information of KGs has not been fully leveraged. In this paper, we propose a novel framework based on Relation-aware Graph Attention Networks to capture the interactions between entities and relations. Our framework adopts the self-attention mechanism to spread entity information to the relations and then aggregate relation information back to entities. Furthermore, we propose a global alignment algorithm to make one-to-one entity alignments with a fine-grained similarity matrix. Experiments on three real-world cross-lingual datasets show that our framework outperforms the state-of-the-art methods.

cs.CL↗

arXiv2021

Tunable ferroelectricity in hBN intercalated twisted double-layer graphene

Van der Waals (vdW) assembly of two-dimensional materials has been long recognized as a powerful tool to create unique systems with properties that cannot be found in natural compounds. However, among the variety of vdW heterostructures and their various properties, only a few have revealed metallic and ferroelectric behaviour signatures. Here we show ferroelectric semimetal made of double-gated double-layer graphene separated by an atomically thin crystal of hexagonal boron nitride, which demonstrating high room temperature mobility of the order of 10 m$^2$V$^{-1}$s$^{-1}$ and exhibits robust ambipolar switching in response to the external electric field. The observed hysteresis is tunable, reversible and persists above room temperature. Our fabrication method expands the family of ferroelectric vdW compounds and offers a route for developing novel phase-changing devices.

cond-mat.mes-hall↗

arXiv2021

Millimeter Wave MIMO based Depth Maps for Wireless Virtual and Augmented Reality

Augmented and virtual reality systems (AR/VR) are rapidly becoming key components of the wireless landscape. For immersive AR/VR experience, these devices should be able to construct accurate depth perception of the surrounding environment. Current AR/VR devices rely heavily on using RGB-D depth cameras to achieve this goal. The performance of these depth cameras, however, has clear limitations in several scenarios, such as the cases with shiny objects, dark surfaces, and abrupt color transition among other limitations. In this paper, we propose a novel solution for AR/VR depth map construction using mmWave MIMO communication transceivers. This is motivated by the deployment of advanced mmWave communication systems in future AR/VR devices for meeting the high data rate demands and by the interesting propagation characteristics of mmWave signals. Accounting for the constraints on these systems, we develop a comprehensive framework for constructing accurate and high-resolution depth maps using mmWave systems. In this framework, we developed new sensing beamforming codebook approaches that are specific for the depth map construction objective. Using these codebooks, and leveraging tools from successive interference cancellation, we develop a joint beam processing approach that can construct high-resolution depth maps using practical mmWave antenna arrays. Extensive simulation results highlight the potential of the proposed solution in building accurate depth maps. Further, these simulations show the promising gains of mmWave based depth perception compared to RGB-based approaches in several important use cases.

eess.SP↗

arXiv2021

Self-paced and self-consistent co-training for semi-supervised image segmentation

Deep co-training has recently been proposed as an effective approach for image segmentation when annotated data is scarce. In this paper, we improve existing approaches for semi-supervised segmentation with a self-paced and self-consistent co-training method. To help distillate information from unlabeled images, we first design a self-paced learning strategy for co-training that lets jointly-trained neural networks focus on easier-to-segment regions first, and then gradually consider harder ones.This is achieved via an end-to-end differentiable loss inthe form of a generalized Jensen Shannon Divergence(JSD). Moreover, to encourage predictions from different networks to be both consistent and confident, we enhance this generalized JSD loss with an uncertainty regularizer based on entropy. The robustness of individual models is further improved using a self-ensembling loss that enforces their prediction to be consistent across different training iterations. We demonstrate the potential of our method on three challenging image segmentation problems with different image modalities, using small fraction of labeled data. Results show clear advantages in terms of performance compared to the standard co-training baselines and recently proposed state-of-the-art approaches for semi-supervised segmentation

cs.CV↗

arXiv2021

Ambient Backscatter-Assisted Wireless-Powered Relaying

Internet-of-Things (IoT) is featured with low-power communications among a massive number of ubiquitously-deployed and energy-constrained electronics, e.g., sensors and actuators. To cope with the demand, wireless-powered cooperative relaying emerges as a promising communication paradigm to extend data transmission coverage and solve energy scarcity for the IoT devices. In this paper, we propose a novel hybrid relaying strategy by combining wireless-powered communication and ambient backscattering functions to improve applicability and performance of data transfer. In particular, the hybrid relay can harvest energy from radio frequency (RF) signals and use the energy for active transmission. Alternatively, the hybrid relay can choose to perform ambient backscattering of incident RF signals for passive transmission. To efficiently utilize the ambient RF resource, we design mode selection protocols to coordinate between the active and passive relaying in circumstances with and without instantaneous channel gain. With different mode selection protocols, we characterize the success probability and ergodic capacity of a dual-hop relaying system with the hybrid relay in the field of randomly located ambient transmitters. The analytical and the numerical results demonstrate the effectiveness of the mode selection protocols in adapting the hybrid relaying into the network environment and reveal the impacts of system parameters on the performance gain of the hybrid relaying. As applications of our analytical framework which is computationally tractable, we formulate optimization problems based on the derived expressions to optimize the system parameters with different objectives. The optimal solutions exhibit a tradeoff between the maximum energy efficiency and target success probability.

cs.IT↗

arXiv2021

Few-shot Image Classification with Multi-Facet Prototypes

The aim of few-shot learning (FSL) is to learn how to recognize image categories from a small number of training examples. A central challenge is that the available training examples are normally insufficient to determine which visual features are most characteristic of the considered categories. To address this challenge, we organize these visual features into facets, which intuitively group features of the same kind (e.g. features that are relevant to shape, color, or texture). This is motivated from the assumption that (i) the importance of each facet differs from category to category and (ii) it is possible to predict facet importance from a pre-trained embedding of the category names. In particular, we propose an adaptive similarity measure, relying on predicted facet importance weights for a given set of categories. This measure can be used in combination with a wide array of existing metric-based methods. Experiments on miniImageNet and CUB show that our approach improves the state-of-the-art in metric-based FSL.

cs.CV↗