arXiv subjects

Li Zhang

Publications and source records attributed to Li Zhang.

At least 703 records · Page 39Linked to original sources

arXiv2021

Multi-Level Transfer Learning from Near-Field to Far-Field Speaker Verification

In far-field speaker verification, the performance of speaker embeddings is susceptible to degradation when there is a mismatch between the conditions of enrollment and test speech. To solve this problem, we propose the feature-level and instance-level transfer learning in the teacher-student framework to learn a domain-invariant embedding space. For the feature-level knowledge transfer, we develop the contrastive loss to transfer knowledge from teacher model to student model, which can not only decrease the intra-class distance, but also enlarge the inter-class distance. Moreover, we propose the instance-level pairwise distance transfer method to force the student model to preserve pairwise instances distance from the well optimized embedding space of the teacher model. On FFSVC 2020 evaluation set, our EER on Full-eval trials is relatively reduced by 13.9% compared with the fusion system result on Partial-eval trials of Task2. On Task1, compared with the winner's DenseNet result on Partial-eval trials, our minDCF on Full-eval trials is relatively reduced by 6.3%. On Task3, the EER and minDCF of our proposed method on Full-eval trials are very close to the result of the fusion system on Partial-eval trials. Our results also outperform other competitive domain adaptation methods.

cs.SD↗

arXiv2021

Demonstration of topological wireless power transfer

Recent advances in non-radiative wireless power transfer (WPT) technique essentially relying on magnetic resonance and near-field coupling have successfully enabled a wide range of applications. However, WPT systems based on double resonators are severely limited to short- or mid-range distance, due to the deteriorating efficiency and power with long transfer distance. WPT systems based on multi-relay resonators can overcome this problem, which, however, suffer from sensitivity to perturbations and fabrication imperfections. Here, we experimentally demonstrate a concept of topological wireless power transfer (TWPT), where energy is transferred efficiently via the near-field coupling between two topological edge states localized at the ends of a one-dimensional radiowave topological insulator. Such a TWPT system can be modelled as a parity-time-symmetric Su-Schrieffer-Heeger (SSH) chain with complex boundary potentials. Besides, the coil configurations are judiciously designed, which significantly suppress the unwanted cross-couplings between nonadjacent coils that could break the chiral symmetry of the SSH chain. By tuning the inter- and intra-cell coupling strengths, we theoretically and experimentally demonstrate high energy transfer efficiency near the exceptional point of the topological edge states, even in the presence of disorder. The combination of topological metamaterials, non-Hermitian physics, and WPT techniques could promise a variety of robust, efficient WPT applications over long distances in electronics, transportation, and industry.

physics.app-ph↗

arXiv2021

Composite Localization for Human Pose Estimation

The existing human pose estimation methods are confronted with inaccurate long-distance regression or high computational cost due to the complex learning objectives. This work proposes a novel deep learning framework for human pose estimation called composite localization to divide the complex learning objective into two simpler ones: a sparse heatmap to find the keypoint's approximate location and two short-distance offsetmaps to obtain its final precise coordinates. To realize the framework, we construct two types of composite localization networks: CLNet-ResNet and CLNet-Hourglass. We evaluate the networks on three benchmark datasets, including the Leeds Sports Pose dataset, the MPII Human Pose dataset, and the COCO keypoints detection dataset. The experimental results show that our CLNet-ResNet50 outperforms SimpleBaseline by 1.14% with about 1/2 GFLOPs. Our CLNet-Hourglass outperforms the original stacked-hourglass by 4.45% on COCO.

cs.CV↗

arXiv2021

How to trust unlabeled data? Instance Credibility Inference for Few-Shot Learning

Deep learning based models have excelled in many computer vision tasks and appear to surpass humans' performance. However, these models require an avalanche of expensive human labeled training data and many iterations to train their large number of parameters. This severely limits their scalability to the real-world long-tail distributed categories, some of which are with a large number of instances, but with only a few manually annotated. Learning from such extremely limited labeled examples is known as Few-shot learning (FSL). Different to prior arts that leverage meta-learning or data augmentation strategies to alleviate this extremely data-scarce problem, this paper presents a statistical approach, dubbed Instance Credibility Inference (ICI) to exploit the support of unlabeled instances for few-shot visual recognition. Typically, we repurpose the self-taught learning paradigm to predict pseudo-labels of unlabeled instances with an initial classifier trained from the few shot and then select the most confident ones to augment the training set to re-train the classifier. This is achieved by constructing a (Generalized) Linear Model (LM/GLM) with incidental parameters to model the mapping from (un-)labeled features to their (pseudo-)labels, in which the sparsity of the incidental parameters indicates the credibility of the corresponding pseudo-labeled instance. We rank the credibility of pseudo-labeled instances along the regularization path of their corresponding incidental parameters, and the most trustworthy pseudo-labeled examples are preserved as the augmented labeled instances. Theoretically, under mild conditions of restricted eigenvalue, irrepresentability, and large error, our approach is guaranteed to collect all the correctly-predicted instances from the noisy pseudo-labeled set.

cs.CV↗

arXiv2021

BasisNet: Two-stage Model Synthesis for Efficient Inference

In this work, we present BasisNet which combines recent advancements in efficient neural network architectures, conditional computation, and early termination in a simple new form. Our approach incorporates a lightweight model to preview the input and generate input-dependent combination coefficients, which later controls the synthesis of a more accurate specialist model to make final prediction. The two-stage model synthesis strategy can be applied to any network architectures and both stages are jointly trained. We also show that proper training recipes are critical for increasing generalizability for such high capacity neural networks. On ImageNet classification benchmark, our BasisNet with MobileNets as backbone demonstrated clear advantage on accuracy-efficiency trade-off over several strong baselines. Specifically, BasisNet-MobileNetV3 obtained 80.3% top-1 accuracy with only 290M Multiply-Add operations, halving the computational cost of previous state-of-the-art without sacrificing accuracy. With early termination, the average cost can be further reduced to 198M MAdds while maintaining accuracy of 80.0% on ImageNet.

cs.CV↗

arXiv2021

Hybrid Template Canonical Correlation Analysis Method for Enhancing SSVEP Recognition under data-limited Condition

In this study, an advanced CCA-based algorithn called hybrid template canonical correlation analysis (HTCCA) was proposed to improve the performance of brain-computer interface (BCI) based on steady state visual evoked potential (SSVEP) uuder data-linited condition. The HTCCA method combines the training data from several subjects to construct SSVEP templates. The experinental results evaluated on two public benchmark datasets showed that the proposed method outperforms the compared methods in both detection accuracy and information transfer rate when the number of tuials is small.Considering that user-friendly experience will become a key factor for BCI in practical application, it is very necessary to develop effective methods based on limited EEG samples. This study demonstrates that the proposed method has great potential in the application of SSVEP-based brain-computer interfaces.

eess.SP↗

arXiv2021

Prediction of clinical tremor severity using Rank Consistent Ordinal Regression

Tremor is a key diagnostic feature of Parkinson's Disease (PD), Essential Tremor (ET), and other central nervous system (CNS) disorders. Clinicians or trained raters assess tremor severity with TETRAS scores by observing patients. Lacking quantitative measures, inter- or intra- observer variabilities are almost inevitable as the distinction between adjacent tremor scores is subtle. Moreover, clinician assessments also require patient visits, which limits the frequency of disease progress evaluation. Therefore it is beneficial to develop an automated assessment that can be performed remotely and repeatably at patients' convenience for continuous monitoring. In this work, we proposed to train a deep neural network (DNN) with rank-consistent ordinal regression using 276 clinical videos from 36 essential tremor patients. The videos are coupled with clinician assessed TETRAS scores, which are used as ground truth labels to train the DNN. To tackle the challenge of limited training data, optical flows are used to eliminate irrelevant background and statistic objects from RGB frames. In addition to optical flows, transfer learning is also applied to leverage pre-trained network weights from a related task of tremor frequency estimate. The approach was evaluated by splitting the clinical videos into training (67%) and testing sets (0.33%). The mean absolute error on TETRAS score of the testing results is 0.45, indicating that most of the errors were from the mismatch of adjacent labels, which is expected and acceptable. The model predications also agree well with clinical ratings. This model is further applied to smart phone videos collected from a PD patient who has an implanted device to turn "On" or "Off" tremor. The model outputs were consistent with the patient tremor states. The results demonstrate that our trained model can be used as a means to assess and track tremor severity.

cs.CV↗

arXiv2021

Delving into Data: Effectively Substitute Training for Black-box Attack

Deep models have shown their vulnerability when processing adversarial samples. As for the black-box attack, without access to the architecture and weights of the attacked model, training a substitute model for adversarial attacks has attracted wide attention. Previous substitute training approaches focus on stealing the knowledge of the target model based on real training data or synthetic data, without exploring what kind of data can further improve the transferability between the substitute and target models. In this paper, we propose a novel perspective substitute training that focuses on designing the distribution of data used in the knowledge stealing process. More specifically, a diverse data generation module is proposed to synthesize large-scale data with wide distribution. And adversarial substitute training strategy is introduced to focus on the data distributed near the decision boundary. The combination of these two modules can further boost the consistency of the substitute model and target model, which greatly improves the effectiveness of adversarial attack. Extensive experiments demonstrate the efficacy of our method against state-of-the-art competitors under non-target and target attack settings. Detailed visualization and analysis are also provided to help understand the advantage of our method.

cs.CV↗

arXiv2021

Optimize Neural Fictitious Self-Play in Regret Minimization Thinking

Optimization of deep learning algorithms to approach Nash Equilibrium remains a significant problem in imperfect information games, e.g. StarCraft and poker. Neural Fictitious Self-Play (NFSP) has provided an effective way to learn approximate Nash Equilibrium without prior domain knowledge in imperfect information games. However, optimality gap was left as an optimization problem of NFSP and by solving the problem, the performance of NFSP could be improved. In this study, focusing on the optimality gap of NFSP, we have proposed a new method replacing NFSP's best response computation with regret matching method. The new algorithm can make the optimality gap converge to zero as it iterates, thus converge faster than original NFSP. We have conduct experiments on three typical environments of perfect-information games and imperfect information games in OpenSpiel and all showed that our new algorithm performances better than original NFSP.

cs.AI↗

arXiv2021

MoViNets: Mobile Video Networks for Efficient Video Recognition

We present Mobile Video Networks (MoViNets), a family of computation and memory efficient video networks that can operate on streaming video for online inference. 3D convolutional neural networks (CNNs) are accurate at video recognition but require large computation and memory budgets and do not support online inference, making them difficult to work on mobile devices. We propose a three-step approach to improve computational efficiency while substantially reducing the peak memory usage of 3D CNNs. First, we design a video network search space and employ neural architecture search to generate efficient and diverse 3D CNN architectures. Second, we introduce the Stream Buffer technique that decouples memory from video clip duration, allowing 3D CNNs to embed arbitrary-length streaming video sequences for both training and inference with a small constant memory footprint. Third, we propose a simple ensembling technique to improve accuracy further without sacrificing efficiency. These three progressive techniques allow MoViNets to achieve state-of-the-art accuracy and efficiency on the Kinetics, Moments in Time, and Charades video action recognition datasets. For instance, MoViNet-A5-Stream achieves the same accuracy as X3D-XL on Kinetics 600 while requiring 80% fewer FLOPs and 65% less memory. Code will be made available at https://github.com/tensorflow/models/tree/master/official/vision.

cs.CV↗

arXiv2021

Joint Optimization of Preamble Selection and Access Barring for MTC with Correlated Device Activities

Most existing works on random access for machine-type communication (MTC) assume independent device activities. However, in several Internet-of-Things (IoT) applications, device activities are driven by events and hence may be correlated. This paper investigates the joint optimization of preamble selection and access barring for correlated device activities. We adopt a random access scheme with general random preamble selection parameterized by the preamble selection distributions of all devices and an access barring scheme parameterized by the access barring factor, to maximally exploit correlated device activities for improving the average throughput. First, we formulate the average throughput maximization problem with respect to the preamble selection distributions and the access barring factor. It is a challenging nonconvex problem. We characterize an optimality property of the problem. Then, we develop two iterative algorithms to obtain a stationary point and a low-complexity solution respectively by using the block coordinate descend (BCD) method. Numerical results show that the two proposed solutions achieve significant gains over existing schemes, demonstrating the significance of exploiting correlation of device activities in improving the average throughput. Numerical results also show that compared to the stationary point, the low-complexity solution achieves a similar average throughput with much lower computational complexity, demonstrating the effectiveness of the low-complexity solution.

cs.IT↗

arXiv2021

Large evanescently-induced Brillouin scattering at the surrounding of a nanofibre

Brillouin scattering has been widely exploited for advanced photonics functionalities such as microwave photonics, signal processing, sensing, lasing, and more recently in micro- and nano-photonic waveguides. So far, all the works have focused on the opto-acoustic interaction driven from the core region of micro- and nano-waveguides. Here we observe, for the first time, an efficient Brillouin scattering generated by an evanescent field nearby a sub-wavelength waveguide embedded in a pressurised gas cell, with a maximum gain coefficient of $18.90 \pm 0.17$ m$^{-1}$W$^{-1}$. This gain is 11 times larger than the highest Brillouin gain obtained in a hollow-core fibre and 79 times larger than in a standard single-mode fibre. The realisation of strong free-space Brillouin scattering from a waveguide benefits from the flexibility of confined light while providing a direct access to the opto-acoustic interaction, as required in free-space optoacoustics such as Brillouin spectroscopy and microscopy. Therefore, our work creates an important bridge between Brillouin scattering in waveguides, Brillouin spectroscopy and microscopy, and opens new avenues in light-sound interactions, optomechanics, sensing, lasing and imaging.

physics.optics↗

arXiv2021

BEFD: Boundary Enhancement and Feature Denoising for Vessel Segmentation

Blood vessel segmentation is crucial for many diagnostic and research applications. In recent years, CNN-based models have leaded to breakthroughs in the task of segmentation, however, such methods usually lose high-frequency information like object boundaries and subtle structures, which are vital to vessel segmentation. To tackle this issue, we propose Boundary Enhancement and Feature Denoising (BEFD) module to facilitate the network ability of extracting boundary information in semantic segmentation, which can be integrated into arbitrary encoder-decoder architecture in an end-to-end way. By introducing Sobel edge detector, the network is able to acquire additional edge prior, thus enhancing boundary in an unsupervised manner for medical image segmentation. In addition, we also utilize a denoising block to reduce the noise hidden in the low-level features. Experimental results on retinal vessel dataset and angiocarpy dataset demonstrate the superior performance of the new BEFD module.

eess.IV↗

arXiv2021

Three Decades of Deception Techniques in Active Cyber Defense -- Retrospect and Outlook

Deception techniques have been widely seen as a game changer in cyber defense. In this paper, we review representative techniques in honeypots, honeytokens, and moving target defense, spanning from the late 1980s to the year 2021. Techniques from these three domains complement with each other and may be leveraged to build a holistic deception based defense. However, to the best of our knowledge, there has not been a work that provides a systematic retrospect of these three domains all together and investigates their integrated usage for orchestrated deceptions. Our paper aims to fill this gap. By utilizing a tailored cyber kill chain model which can reflect the current threat landscape and a four-layer deception stack, a two-dimensional taxonomy is developed, based on which the deception techniques are classified. The taxonomy literally answers which phases of a cyber attack campaign the techniques can disrupt and which layers of the deception stack they belong to. Cyber defenders may use the taxonomy as a reference to design an organized and comprehensive deception plan, or to prioritize deception efforts for a budget conscious solution. We also discuss two important points for achieving active and resilient cyber defense, namely deception in depth and deception lifecycle, where several notable proposals are illustrated. Finally, some outlooks on future research directions are presented, including dynamic integration of different deception techniques, quantified deception effects and deception operation cost, hardware-supported deception techniques, as well as techniques developed based on better understanding of the human element.

cs.CR↗

arXiv2021

Complementary Evidence Identification in Open-Domain Question Answering

This paper proposes a new problem of complementary evidence identification for open-domain question answering (QA). The problem aims to efficiently find a small set of passages that covers full evidence from multiple aspects as to answer a complex question. To this end, we proposes a method that learns vector representations of passages and models the sufficiency and diversity within the selected set, in addition to the relevance between the question and passages. Our experiments demonstrate that our method considers the dependence within the supporting evidence and significantly improves the accuracy of complementary evidence selection in QA domain.

cs.CL↗

arXiv2021

Hierarchical Road Topology Learning for Urban Map-less Driving

The majority of current approaches in autonomous driving rely on High-Definition (HD) maps which detail the road geometry and surrounding area. Yet, this reliance is one of the obstacles to mass deployment of autonomous vehicles due to poor scalability of such prior maps. In this paper, we tackle the problem of online road map extraction via leveraging the sensory system aboard the vehicle itself. To this end, we design a structured model where a graph representation of the road network is generated in a hierarchical fashion within a fully convolutional network. The method is able to handle complex road topology and does not require a user in the loop.

cs.RO↗

arXiv2021

Depth-conditioned Dynamic Message Propagation for Monocular 3D Object Detection

The objective of this paper is to learn context- and depth-aware feature representation to solve the problem of monocular 3D object detection. We make following contributions: (i) rather than appealing to the complicated pseudo-LiDAR based approach, we propose a depth-conditioned dynamic message propagation (DDMP) network to effectively integrate the multi-scale depth information with the image context;(ii) this is achieved by first adaptively sampling context-aware nodes in the image context and then dynamically predicting hybrid depth-dependent filter weights and affinity matrices for propagating information; (iii) by augmenting a center-aware depth encoding (CDE) task, our method successfully alleviates the inaccurate depth prior; (iv) we thoroughly demonstrate the effectiveness of our proposed approach and show state-of-the-art results among the monocular-based approaches on the KITTI benchmark dataset. Particularly, we rank $1^{st}$ in the highly competitive KITTI monocular 3D object detection track on the submission day (November 16th, 2020). Code and models are released at \url{https://github.com/fudan-zvg/DDMP}

cs.CV↗

arXiv2021

Few-shot Action Recognition with Prototype-centered Attentive Learning

Few-shot action recognition aims to recognize action classes with few training samples. Most existing methods adopt a meta-learning approach with episodic training. In each episode, the few samples in a meta-training task are split into support and query sets. The former is used to build a classifier, which is then evaluated on the latter using a query-centered loss for model updating. There are however two major limitations: lack of data efficiency due to the query-centered only loss design and inability to deal with the support set outlying samples and inter-class distribution overlapping problems. In this paper, we overcome both limitations by proposing a new Prototype-centered Attentive Learning (PAL) model composed of two novel components. First, a prototype-centered contrastive learning loss is introduced to complement the conventional query-centered learning objective, in order to make full use of the limited training samples in each episode. Second, PAL further integrates a hybrid attentive learning mechanism that can minimize the negative impacts of outliers and promote class separation. Extensive experiments on four standard few-shot action benchmarks show that our method clearly outperforms previous state-of-the-art methods, with the improvement particularly significant (10+\%) on the most challenging fine-grained action recognition benchmark.

cs.CV↗