arXiv subjects

Hao Yang

Publications and source records attributed to Hao Yang.

At least 379 records · Page 21Linked to original sources

arXiv2021

Edge Computing for Real-Time Near-Crash Detection for Smart Transportation Applications

Traffic near-crash events serve as critical data sources for various smart transportation applications, such as being surrogate safety measures for traffic safety research and corner case data for automated vehicle testing. However, there are several key challenges for near-crash detection. First, extracting near-crashes from original data sources requires significant computing, communication, and storage resources. Also, existing methods lack efficiency and transferability, which bottlenecks prospective large-scale applications. To this end, this paper leverages the power of edge computing to address these challenges by processing the video streams from existing dashcams onboard in a real-time manner. We design a multi-thread system architecture that operates on edge devices and model the bounding boxes generated by object detection and tracking in linear complexity. The method is insensitive to camera parameters and backward compatible with different vehicles. The edge computing system has been evaluated with recorded videos and real-world tests on two cars and four buses for over ten thousand hours. It filters out irrelevant videos in real-time thereby saving labor cost, processing time, network bandwidth, and data storage. It collects not only event videos but also other valuable data such as road user type, event location, time to collision, vehicle trajectory, vehicle speed, brake switch, and throttle. The experiments demonstrate the promising performance of the system regarding efficiency, accuracy, reliability, and transferability. It is among the first efforts in applying edge computing for real-time traffic video analytics and is expected to benefit multiple sub-fields in smart transportation research and applications.

cs.RO↗

arXiv2021

Event2Graph: Event-driven Bipartite Graph for Multivariate Time-series Anomaly Detection

Modeling inter-dependencies between time-series is the key to achieve high performance in anomaly detection for multivariate time-series data. The de-facto solution to model the dependencies is to feed the data into a recurrent neural network (RNN). However, the fully connected network structure underneath the RNN (either GRU or LSTM) assumes a static and complete dependency graph between time-series, which may not hold in many real-world applications. To alleviate this assumption, we propose a dynamic bipartite graph structure to encode the inter-dependencies between time-series. More concretely, we model time series as one type of nodes, and the time series segments (regarded as event) as another type of nodes, where the edge between two types of nodes describe a temporal pattern occurred on a specific time series at a certain time. Based on this design, relations between time series can be explicitly modelled via dynamic connections to event nodes, and the multivariate time-series anomaly detection problem can be formulated as a self-supervised, edge stream prediction problem in dynamic graphs. We conducted extensive experiments to demonstrate the effectiveness of the design.

cs.LG↗

arXiv2021

The HW-TSC's Offline Speech Translation Systems for IWSLT 2021 Evaluation

This paper describes our work in participation of the IWSLT-2021 offline speech translation task. Our system was built in a cascade form, including a speaker diarization module, an Automatic Speech Recognition (ASR) module and a Machine Translation (MT) module. We directly use the LIUM SpkDiarization tool as the diarization module. The ASR module is trained with three ASR datasets from different sources, by multi-source training, using a modified Transformer encoder. The MT module is pretrained on the large-scale WMT news translation dataset and fine-tuned on the TED corpus. Our method achieves 24.6 BLEU score on the 2021 test set.

cs.CL↗

arXiv2021

Task and Situation Structures for Service Agent Planning

Everyday tasks are characterized by their varieties and variations, and frequently are not clearly specified to service agents. This paper presents a comprehensive approach to enable a service agent to deal with everyday tasks in open, uncontrolled environments. We introduce a generic structure for representing tasks, and another structure for representing situations. Based on the two newly introduced structures, we present a methodology of situation handling that avoids hard-coding domain rules while improving the scalability of real-world task planning systems.

cs.AI↗

arXiv2021

Style-based Point Generator with Adversarial Rendering for Point Cloud Completion

In this paper, we proposed a novel Style-based Point Generator with Adversarial Rendering (SpareNet) for point cloud completion. Firstly, we present the channel-attentive EdgeConv to fully exploit the local structures as well as the global shape in point features. Secondly, we observe that the concatenation manner used by vanilla foldings limits its potential of generating a complex and faithful shape. Enlightened by the success of StyleGAN, we regard the shape feature as style code that modulates the normalization layers during the folding, which considerably enhances its capability. Thirdly, we realize that existing point supervisions, e.g., Chamfer Distance or Earth Mover's Distance, cannot faithfully reflect the perceptual quality of the reconstructed points. To address this, we propose to project the completed points to depth maps with a differentiable renderer and apply adversarial training to advocate the perceptual realism under different viewpoints. Comprehensive experiments on ShapeNet and KITTI prove the effectiveness of our method, which achieves state-of-the-art quantitative performance while offering superior visual quality.

cs.CV↗

arXiv2021

Unsupervised Pre-training for Person Re-identification

In this paper, we present a large scale unlabeled person re-identification (Re-ID) dataset "LUPerson" and make the first attempt of performing unsupervised pre-training for improving the generalization ability of the learned person Re-ID feature representation. This is to address the problem that all existing person Re-ID datasets are all of limited scale due to the costly effort required for data annotation. Previous research tries to leverage models pre-trained on ImageNet to mitigate the shortage of person Re-ID data but suffers from the large domain gap between ImageNet and person Re-ID data. LUPerson is an unlabeled dataset of 4M images of over 200K identities, which is 30X larger than the largest existing Re-ID dataset. It also covers a much diverse range of capturing environments (eg, camera settings, scenes, etc.). Based on this dataset, we systematically study the key factors for learning Re-ID features from two perspectives: data augmentation and contrastive loss. Unsupervised pre-training performed on this large-scale dataset effectively leads to a generic Re-ID feature that can benefit all existing person Re-ID methods. Using our pre-trained model in some basic frameworks, our methods achieve state-of-the-art results without bells and whistles on four widely used Re-ID datasets: CUHK03, Market1501, DukeMTMC, and MSMT17. Our results also show that the performance improvement is more significant on small-scale target datasets or under few-shot setting.

cs.CV↗

arXiv2021

Integrating Subgraph-aware Relation and DirectionReasoning for Question Answering

Question Answering (QA) models over Knowledge Bases (KBs) are capable of providing more precise answers by utilizing relation information among entities. Although effective, most of these models solely rely on fixed relation representations to obtain answers for different question-related KB subgraphs. Hence, the rich structured information of these subgraphs may be overlooked by the relation representation vectors. Meanwhile, the direction information of reasoning, which has been proven effective for the answer prediction on graphs, has not been fully explored in existing work. To address these challenges, we propose a novel neural model, Relation-updated Direction-guided Answer Selector (RDAS), which converts relations in each subgraph to additional nodes to learn structure information. Additionally, we utilize direction information to enhance the reasoning ability. Experimental results show that our model yields substantial improvements on two widely used datasets.

cs.CL↗

arXiv2021

Learning from Noisy Labels via Dynamic Loss Thresholding

Numerous researches have proved that deep neural networks (DNNs) can fit everything in the end even given data with noisy labels, and result in poor generalization performance. However, recent studies suggest that DNNs tend to gradually memorize the data, moving from correct data to mislabeled data. Inspired by this finding, we propose a novel method named Dynamic Loss Thresholding (DLT). During the training process, DLT records the loss value of each sample and calculates dynamic loss thresholds. Specifically, DLT compares the loss value of each sample with the current loss threshold. Samples with smaller losses can be considered as clean samples with higher probability and vice versa. Then, DLT discards the potentially corrupted labels and further leverages supervised learning techniques. Experiments on CIFAR-10/100 and Clothing1M demonstrate substantial improvements over recent state-of-the-art methods. In addition, we investigate two real-world problems for the first time. Firstly, we propose a novel approach to estimate the noise rates of datasets based on the loss difference between the early and late training stages of DNNs. Secondly, we explore the effect of hard samples (which are difficult to be distinguished) on the process of learning from noisy labels.

cs.LG↗

arXiv2021

Gaussian State-Based Quantum Illumination with Simple Photodetection

Proofs of the quantum advantage available in imaging or detecting objects under quantum illumination can rely on optimal measurements without specifying what they are. We use the continuous-variable Gaussian quantum information formalism to show that quantum illumination is better for object detection compared with coherent states of the same mean photon number, even for simple direct photodetection. The advantage persists if signal energy and object reflectivity are low and background thermal noise is high. The advantage is even greater if we match signal beam detection probabilities rather than mean photon number. We perform all calculations with thermal states, even for non-Gaussian conditioned states with negative Wigner functions. We simulate repeated detection using a Monte Carlo process that clearly shows the advantages obtainable.

quant-ph↗

arXiv2021

A coarse-to-fine framework for unsupervised multi-contrast MR image deformable registration with dual consistency constraint

Multi-contrast magnetic resonance (MR) image registration is useful in the clinic to achieve fast and accurate imaging-based disease diagnosis and treatment planning. Nevertheless, the efficiency and performance of the existing registration algorithms can still be improved. In this paper, we propose a novel unsupervised learning-based framework to achieve accurate and efficient multi-contrast MR image registrations. Specifically, an end-to-end coarse-to-fine network architecture consisting of affine and deformable transformations is designed to improve the robustness and achieve end-to-end registration. Furthermore, a dual consistency constraint and a new prior knowledge-based loss function are developed to enhance the registration performances. The proposed method has been evaluated on a clinical dataset containing 555 cases, and encouraging performances have been achieved. Compared to the commonly utilized registration methods, including VoxelMorph, SyN, and LT-Net, the proposed method achieves better registration performance with a Dice score of 0.8397 in identifying stroke lesions. With regards to the registration speed, our method is about 10 times faster than the most competitive method of SyN (Affine) when testing on a CPU. Moreover, we prove that our method can still perform well on more challenging tasks with lacking scanning information data, showing high robustness for the clinical application.

eess.IV↗

arXiv2021

Large-Scale Training System for 100-Million Classification at Alibaba

In the last decades, extreme classification has become an essential topic for deep learning. It has achieved great success in many areas, especially in computer vision and natural language processing (NLP). However, it is very challenging to train a deep model with millions of classes due to the memory and computation explosion in the last output layer. In this paper, we propose a large-scale training system to address these challenges. First, we build a hybrid parallel training framework to make the training process feasible. Second, we propose a novel softmax variation named KNN softmax, which reduces both the GPU memory consumption and computation costs and improves the throughput of training. Then, to eliminate the communication overhead, we propose a new overlapping pipeline and a gradient sparsification method. Furthermore, we design a fast continuous convergence strategy to reduce total training iterations by adaptively adjusting learning rate and updating model parameters. With the help of all the proposed methods, we gain 3.9$\times$ throughput of our training system and reduce almost 60\% of training iterations. The experimental results show that using an in-house 256 GPUs cluster, we could train a classifier of 100 million classes on Alibaba Retail Product Dataset in about five days while achieving a comparable accuracy with the naive softmax training process.

cs.LG↗

arXiv2020

Beating Attackers At Their Own Games: Adversarial Example Detection Using Adversarial Gradient Directions

Adversarial examples are input examples that are specifically crafted to deceive machine learning classifiers. State-of-the-art adversarial example detection methods characterize an input example as adversarial either by quantifying the magnitude of feature variations under multiple perturbations or by measuring its distance from estimated benign example distribution. Instead of using such metrics, the proposed method is based on the observation that the directions of adversarial gradients when crafting (new) adversarial examples play a key role in characterizing the adversarial space. Compared to detection methods that use multiple perturbations, the proposed method is efficient as it only applies a single random perturbation on the input example. Experiments conducted on two different databases, CIFAR-10 and ImageNet, show that the proposed detection method achieves, respectively, 97.9% and 98.6% AUC-ROC (on average) on five different adversarial attacks, and outperforms multiple state-of-the-art detection methods. Results demonstrate the effectiveness of using adversarial gradient directions for adversarial example detection.

cs.CV↗

arXiv2020

Multi-Task Incremental Learning for Object Detection

Multi-task learns multiple tasks, while sharing knowledge and computation among them. However, it suffers from catastrophic forgetting of previous knowledge when learned incrementally without access to the old data. Most existing object detectors are domain-specific and static, while some are learned incrementally but only within a single domain. Training an object detector incrementally across various domains has rarely been explored. In this work, we propose three incremental learning scenarios across various domains and categories for object detection. To mitigate catastrophic forgetting, attentive feature distillation is proposed to leverages both bottom-up and top-down attentions to extract important information for distillation. We then systematically analyze the proposed distillation method in different scenarios. We find out that, contrary to common understanding, domain gaps have smaller negative impact on incremental detection, while category differences are problematic. For the difficult cases, where the domain gaps and especially category differences are large, we explore three different exemplar sampling methods and show the proposed adaptive sampling method is effective to select diverse and informative samples from entire datasets, to further prevent forgetting. Experimental results show that we achieve the significant improvement in three different scenarios across seven object detection benchmark datasets.

cs.CV↗

arXiv2020

FaceShifter: Towards High Fidelity And Occlusion Aware Face Swapping

In this work, we propose a novel two-stage framework, called FaceShifter, for high fidelity and occlusion aware face swapping. Unlike many existing face swapping works that leverage only limited information from the target image when synthesizing the swapped face, our framework, in its first stage, generates the swapped face in high-fidelity by exploiting and integrating the target attributes thoroughly and adaptively. We propose a novel attributes encoder for extracting multi-level target face attributes, and a new generator with carefully designed Adaptive Attentional Denormalization (AAD) layers to adaptively integrate the identity and the attributes for face synthesis. To address the challenging facial occlusions, we append a second stage consisting of a novel Heuristic Error Acknowledging Refinement Network (HEAR-Net). It is trained to recover anomaly regions in a self-supervised way without any manual annotations. Extensive experiments on wild faces demonstrate that our face swapping results are not only considerably more perceptually appealing, but also better identity preserving in comparison to other state-of-the-art methods.

cs.CV↗

arXiv2020

Application of light diffraction theory to qualify the downstream light field modulation property of mitigated KDP crystals

Micro-milling can effectively remove laser damage sites on a KDP (potassium dihydrogen phosphate) surface and then improve the laser damage resistance of the components. However, the repaired KDP surface could cause light propagating turbulence and downstream light intensification with the potential risk to damage downstream optics. In order to analyze the downstream light field modulation caused by Gaussian mitigation pits on KDP crystals, a computational model of the downstream light diffraction based on the angular spectrum theory and the Gaussian repair contour is established. The results show that the phase offset caused by the repaired surface produces a large light field modulation near the rear KDP surface. The modulation generated in the whole downstream light field is greater than that caused by the amplitude change. Therefore, the phase characteristics of the outgoing light could be suggested as a vital research topic for future research on the downstream light field modulation caused by mitigation contours. Significantly, the experimental results on the downstream light intensity distribution have good agreement with the simulation ones, which proves the validity of the established downstream light diffraction model. The phase characterization of the outgoing light is proposed as an evaluation tool in the repair of KDP crystals. The developed analytical method and numerical discrete algorithm could be also applicable in qualifying the repair quality of other optical components applied in high-power laser systems.

physics.optics↗

arXiv2020

NLO QCD corrections to $B_c$-pair production in photon-photon collision

The $B_c$ meson pair, including pairs of both pseudoscalar states and vector states, productions in high energy photon-photon interaction are investigated at the next-to-leading order (NLO) accuracy in the nonrelativistic quantum chromodynamics (NRQCD) factorization formalism. The corresponding cross sections at the future $e^+e^-$ colliders with $\sqrt{s}=250$ GeV and $500$ GeV are evaluated. Numerical result indicates that the inclusion of the NLO corrections shall greatly suppress the scale dependence and enhance the prediction reliability. In addition to the phenomenological meaning, the NLO QCD calculation of this process subjects to certain technical issues, which are elucidated in details and might be applicable to other relevant investigations.

hep-ph↗

arXiv2020

Category-Specific CNN for Visual-aware CTR Prediction at JD.com

As one of the largest B2C e-commerce platforms in China, JD com also powers a leading advertising system, serving millions of advertisers with fingertip connection to hundreds of millions of customers. In our system, as well as most e-commerce scenarios, ads are displayed with images.This makes visual-aware Click Through Rate (CTR) prediction of crucial importance to both business effectiveness and user experience. Existing algorithms usually extract visual features using off-the-shelf Convolutional Neural Networks (CNNs) and late fuse the visual and non-visual features for the finally predicted CTR. Despite being extensively studied, this field still face two key challenges. First, although encouraging progress has been made in offline studies, applying CNNs in real systems remains non-trivial, due to the strict requirements for efficient end-to-end training and low-latency online serving. Second, the off-the-shelf CNNs and late fusion architectures are suboptimal. Specifically, off-the-shelf CNNs were designed for classification thus never take categories as input features. While in e-commerce, categories are precisely labeled and contain abundant visual priors that will help the visual modeling. Unaware of the ad category, these CNNs may extract some unnecessary category-unrelated features, wasting CNN's limited expression ability. To overcome the two challenges, we propose Category-specific CNN (CSCNN) specially for CTR prediction. CSCNN early incorporates the category knowledge with a light-weighted attention-module on each convolutional layer. This enables CSCNN to extract expressive category-specific visual patterns that benefit the CTR prediction. Offline experiments on benchmark and a 10 billion scale real production dataset from JD, together with an Online A/B test show that CSCNN outperforms all compared state-of-the-art algorithms.

cs.LG↗

arXiv2020

NLO QCD corrections to $J/ψ$ pair production in photon-photon collision

We calculate the next-to-leading order (NLO) quantum chromodynamics (QCD) correction to the exclusive process $γ+γ\to J/ψ+J/ψ$ in the framework of non-relativistic QCD (NRQCD) factorization formalism. The cross sections at the SuperKEKB electron-positron collider, as well as at the future colliders, like the Circular Electron Positron Collider (CEPC) and the International Linear Collider (ILC), are evaluated. Numerical result indicates that the process will be hopefully to be seen by the Belle II detector within the next decade.

hep-ph↗