arXiv subjects

Jing Yang

Publications and source records attributed to Jing Yang.

At least 289 records · Page 16Linked to original sources

arXiv2022

Few-shot Learning for Multi-modal Social Media Event Filtering

Social media has become an important data source for event analysis. When collecting this type of data, most contain no useful information to a target event. Thus, it is essential to filter out those noisy data at the earliest opportunity for a human expert to perform further inspection. Most existing solutions for event filtering rely on fully supervised methods for training. However, in many real-world scenarios, having access to large number of labeled samples is not possible. To deal with a few labeled sample training problem for event filtering, we propose a graph-based few-shot learning pipeline. We also release the Brazilian Protest Dataset to test our method. To the best of our knowledge, this dataset is the first of its kind in event filtering that focuses on protests in multi-modal social media data, with most of the text in Portuguese. Our experimental results show that our proposed pipeline has comparable performance with only a few labeled samples (60) compared with a fully labeled dataset (3100). To facilitate the research community, we make our dataset and code available at https://github.com/jdnascim/7Set-AL.

cs.LG↗

arXiv2022

FAN-Trans: Online Knowledge Distillation for Facial Action Unit Detection

Due to its importance in facial behaviour analysis, facial action unit (AU) detection has attracted increasing attention from the research community. Leveraging the online knowledge distillation framework, we propose the ``FANTrans" method for AU detection. Our model consists of a hybrid network of convolution and transformer blocks to learn per-AU features and to model AU co-occurrences. The model uses a pre-trained face alignment network as the feature extractor. After further transformation by a small learnable add-on convolutional subnet, the per-AU features are fed into transformer blocks to enhance their representation. As multiple AUs often appear together, we propose a learnable attention drop mechanism in the transformer block to learn the correlation between the features for different AUs. We also design a classifier that predicts AU presence by considering all AUs' features, to explicitly capture label dependencies. Finally, we make the attempt of adapting online knowledge distillation in the training stage for this task, further improving the model's performance. Experiments on the BP4D and DISFA datasets demonstrating the effectiveness of proposed method.

cs.CV↗

arXiv2022

Online Learning with Diverse User Preferences

In this paper, we investigate the impact of diverse user preference on learning under the stochastic multi-armed bandit (MAB) framework. We aim to show that when the user preferences are sufficiently diverse and each arm can be optimal for certain users, the O(log T) regret incurred by exploring the sub-optimal arms under the standard stochastic MAB setting can be reduced to a constant. Our intuition is that to achieve sub-linear regret, the number of times an optimal arm being pulled should scale linearly in time; when all arms are optimal for certain users and pulled frequently, the estimated arm statistics can quickly converge to their true values, thus reducing the need of exploration dramatically. We cast the problem into a stochastic linear bandits model, where both the users preferences and the state of arms are modeled as {independent and identical distributed (i.i.d)} d-dimensional random vectors. After receiving the user preference vector at the beginning of each time slot, the learner pulls an arm and receives a reward as the linear product of the preference vector and the arm state vector. We also assume that the state of the pulled arm is revealed to the learner once its pulled. We propose a Weighted Upper Confidence Bound (W-UCB) algorithm and show that it can achieve a constant regret when the user preferences are sufficiently diverse. The performance of W-UCB under general setups is also completely characterized and validated with synthetic data.

cs.LG↗

arXiv2022

Exploring Graph-aware Multi-View Fusion for Rumor Detection on Social Media

Automatic detecting rumors on social media has become a challenging task. Previous studies focus on learning indicative clues from conversation threads for identifying rumorous information. However, these methods only model rumorous conversation threads from various views but fail to fuse multi-view features very well. In this paper, we propose a novel multi-view fusion framework for rumor representation learning and classification. It encodes the multiple views based on Graph Convolutional Networks (GCN), and leverages Convolutional Neural Networks (CNN) to capture the consistent and complementary information among all views and fuse them together. Experimental results on two public datasets demonstrate that our method outperforms state-of-the-art approaches.

cs.CL↗

arXiv2022

Viscous cosmology in $f(T)$ gravity

We propose a new model for the viscosity of cosmic matters, which can be applied to different epochs of the universe. Using this model, we include the bulk viscosities as practical corrections to the perfect fluid models of the baryonic and dark matters since the material fluids in the real world may have viscosities due to thermodynamics. Such inclusion is put to the test within the framework of $f(T)$ gravity that is proved to be successful in describing the cosmic acceleration, where $T$ denotes the torsion scalar. We perform an observational fit to our model and constrain the cosmological and model parameters by using various latest cosmological datasets. Based on the fitting result, we discuss several cosmological implications including the dissipation of matters, the evolutionary history of the universe, $f(T)$ modification as an effective dark energy, and the Hubble tension problem. The corresponding findings are (i) The late time dissipation will make the density parameters of the matters vanish in the finite future. Moreover, the density ratio between the baryonic and dark matters will change over time. (ii) The radiation dominating era, matter dominating era and the accelerating era can be recovered and the model can successfully describe the known history of the universe. (iii) The $f(T)$ modification is the main drive of the acceleration expansion and currently mimics a phantom-like dark energy. But the universe will eventually enter a de Sitter expansion phase. (iv) The Hubble tension between local and global observations can be significantly alleviated in our model.

gr-qc↗

arXiv2022

Existence and local uniqueness of multi-peak solutions for the Chern-Simons-Schrödinger system

In the present paper, we consider the Chern-Simons-Schrödinger system \begin{equation} \left\{ \begin{aligned} &-\varepsilon^{2}Δu+V(x)u+(A_{0}+A_{1}^{2}+A_{2}^{2})u=|u|^{p-2}u,\,\,\,\,x\in \mathbb{R}^2,\\ &\partial_1 A_0 = A_2 u^2,\ \partial_{2}A_{0}=-A_{1}u^{2},\\ &\partial_{1}A_{2}-\partial_{2}A_{1}=-\frac{1}{2}|u|^{2},\ \partial_{1}A_{1}+\partial_{2}A_{2}=0,\\ \end{aligned} \right. \end{equation} where $p>2,$ $\varepsilon>0$ is a parameter and $V:\mathbb{R}^{2}\rightarrow\mathbb{R}$ is a bounded continuous function. Under some mild assumptions on $V(x)$, we show the existence and local uniqueness of positive multi-peak solutions. Our methods mainly use the finite dimensional reduction method, various local Pohozaev identities, blow-up analysis and the maximum principle. Because of the nonlocal terms involved by $A_{0},A_{1}$ and $A_{2},$ we have to obtain a series of new and technical estimates.

math.AP↗

arXiv2022

Span-based joint entity and relation extraction augmented with sequence tagging mechanism

Span-based joint extraction simultaneously conducts named entity recognition (NER) and relation extraction (RE) in text span form. However, since previous span-based models rely on span-level classifications, they cannot benefit from token-level label information, which has been proven advantageous for the task. In this paper, we propose a Sequence Tagging augmented Span-based Network (STSN), a span-based joint model that can make use of token-level label information. In STSN, we construct a core neural architecture by deep stacking multiple attention layers, each of which consists of three basic attention units. On the one hand, the core architecture enables our model to learn token-level label information via the sequence tagging mechanism and then uses the information in the span-based joint extraction; on the other hand, it establishes a bi-directional information interaction between NER and RE. Experimental results on three benchmark datasets show that STSN consistently outperforms the strongest baselines in terms of F1, creating new state-of-the-art results.

cs.CL↗

arXiv2022

Random Orthogonalization for Federated Learning in Massive MIMO Systems

We propose a novel communication design, termed random orthogonalization, for federated learning (FL) in a massive multiple-input and multiple-output (MIMO) wireless system. The key novelty of random orthogonalization comes from the tight coupling of FL and two unique characteristics of massive MIMO -- channel hardening and favorable propagation. As a result, random orthogonalization can achieve natural over-the-air model aggregation without requiring transmitter side channel state information (CSI) for the uplink phase of FL, while significantly reducing the channel estimation overhead at the receiver. We extend this principle to the downlink communication phase and develop a simple but highly effective model broadcast method for FL. We also relax the massive MIMO assumption by proposing an enhanced random orthogonalization design for both uplink and downlink FL communications, that does not rely on channel hardening or favorable propagation. Theoretical analyses with respect to both communication and machine learning performance are carried out. In particular, an explicit relationship among the convergence rate, the number of clients, and the number of antennas is established. Experimental results validate the effectiveness and efficiency of random orthogonalization for FL in massive MIMO.

cs.IT↗

arXiv2022

Counting Varying Density Crowds Through Density Guided Adaptive Selection CNN and Transformer Estimation

In real-world crowd counting applications, the crowd densities in an image vary greatly. When facing density variation, humans tend to locate and count the targets in low-density regions, and reason the number in high-density regions. We observe that CNN focus on the local information correlation using a fixed-size convolution kernel and the Transformer could effectively extract the semantic crowd information by using the global self-attention mechanism. Thus, CNN could locate and estimate crowds accurately in low-density regions, while it is hard to properly perceive the densities in high-density regions. On the contrary, Transformer has a high reliability in high-density regions, but fails to locate the targets in sparse regions. Neither CNN nor Transformer can well deal with this kind of density variation. To address this problem, we propose a CNN and Transformer Adaptive Selection Network (CTASNet) which can adaptively select the appropriate counting branch for different density regions. Firstly, CTASNet generates the prediction results of CNN and Transformer. Then, considering that CNN/Transformer is appropriate for low/high-density regions, a density guided adaptive selection module is designed to automatically combine the predictions of CNN and Transformer. Moreover, to reduce the influences of annotation noise, we introduce a Correntropy based optimal transport loss. Extensive experiments on four challenging crowd counting datasets have validated the proposed method.

cs.CV↗

arXiv2022

DNA Scaffold Carrier Assisted Protein Translocation Through Solid-State Nanopore

The detection of biomolecules at the single molecular level have important applications in the fields of biosensing and biomedical diagnosis. Solid state nanopore (SS-nanopore) is an effective tool to perform the single molecular detection, due to its unique properties of label-free and less sample consumption. The current SS-nanopore translocations of small biomolecules are driven by the electronic field force, thus easily influenced by the charges, shapes and sizes of the target molecules. Therefore, it remains a great challenge to control biomolecules to translocate through SS-nanopore, particularly for the protein with complex conformations and different charges. Towards this problem, we developed a DNA scaffold carrier assisted strategy to help protein translocation through SS-nanopore, thus facilitating the target protein detection. Taking advantages of a DNA carrier loading effect, the current signal to noise ratios are significantly improved. Our proposed method is sensitive, convenient and can detect real-time sample nanopore translocations. This method opens up a wide range of applications for SS-nanopore detection and promote the development of single molecular detections.

physics.bio-ph↗

arXiv2022

Existence, Local uniqueness and periodicity of bubbling solutions for a critical nonlinear elliptic equation

We revisit the following nonlinear critical elliptic equation \begin{equation*} -Δu+Q(y)u=u^{\frac{N+2}{N-2}},\;\;\; u>0\;\;\;\hbox{ in } \mathbb{R}^N, \end{equation*} where $N\geq 5.$ There seems to be no results about the periodicity of bubbling solutions. Here we try to investigate some related problems. Assuming that $Q(y)$ is periodic in $y_1$ with period 1 and has a local minimum at 0 satisfying $Q(0)=0,$ we prove the existence and local uniqueness of infinitely many bubbling solutions of the problem above. This local uniqueness result implies that some bubbling solutions preserve the symmetry of the potential function $Q(y),$ i.e. the bubbling solution whose blow-up set is $\{(jL,0,...,0):j=0,\pm 1, \pm 2,..., \pm m\}$ must be periodic in $y_{1}$ provided that $L$ is large enough, where $m$ is the number of the bubbles which is large enough but independent of $L.$ Moreover, we also show a non-existence of this bubbling solutions for the problem above if the local minimum of $Q(y)$ does not equal to zero.

math.AP↗

arXiv2022

ArcFace: Additive Angular Margin Loss for Deep Face Recognition

Recently, a popular line of research in face recognition is adopting margins in the well-established softmax loss function to maximize class separability. In this paper, we first introduce an Additive Angular Margin Loss (ArcFace), which not only has a clear geometric interpretation but also significantly enhances the discriminative power. Since ArcFace is susceptible to the massive label noise, we further propose sub-center ArcFace, in which each class contains $K$ sub-centers and training samples only need to be close to any of the $K$ positive sub-centers. Sub-center ArcFace encourages one dominant sub-class that contains the majority of clean faces and non-dominant sub-classes that include hard or noisy faces. Based on this self-propelled isolation, we boost the performance through automatically purifying raw web faces under massive real-world noise. Besides discriminative feature embedding, we also explore the inverse problem, mapping feature vectors to face images. Without training any additional generator or discriminator, the pre-trained ArcFace model can generate identity-preserved face images for both subjects inside and outside the training data only by using the network gradient and Batch Normalization (BN) priors. Extensive experiments demonstrate that ArcFace can enhance the discriminative feature embedding as well as strengthen the generative face synthesis.

cs.CV↗

arXiv2022

Unified BERT for Few-shot Natural Language Understanding

Even as pre-trained language models share a semantic encoder, natural language understanding suffers from a diversity of output schemas. In this paper, we propose UBERT, a unified bidirectional language understanding model based on BERT framework, which can universally model the training objects of different NLU tasks through a biaffine network. Specifically, UBERT encodes prior knowledge from various aspects, uniformly constructing learning representations across multiple NLU tasks, which is conducive to enhancing the ability to capture common semantic understanding. By using the biaffine to model scores pair of the start and end position of the original text, various classification and extraction structures can be converted into a universal, span-decoding approach. Experiments show that UBERT wins the first price in the 2022 AIWIN - World Artificial Intelligence Innovation Competition, Chinese insurance few-shot multi-task track, and realizes the unification of extensive information extraction and linguistic reasoning tasks.

cs.CL↗

arXiv2022

One-Dimensional Quantum Systems with Ground-State of Jastrow Form Are Integrable

The exchange operator formalism (EOF) describes many-body integrable systems using phase-space variables involving an exchange operator that acts on any pair of particles. We establish an equivalence between models described by EOF and the complete infinite family of parent Hamiltonians (PHJ) describing quantum many-body models with ground-states of Jastrow form. This makes it possible to identify the invariants of motion for any model in the PHJ family and establish its integrability, even in the presence of an external potential. Using this construction we establish the integrability of the long-range Lieb-Liniger model, describing bosons in a harmonic trap and subject to contact and Coulomb interactions in one dimension. We give a variety of examples exemplifying the integrability of Hamiltonians in this family.

math-ph↗

arXiv2022

Pre-training strategies and datasets for facial representation learning

What is the best way to learn a universal face representation? Recent work on Deep Learning in the area of face analysis has focused on supervised learning for specific tasks of interest (e.g. face recognition, facial landmark localization etc.) but has overlooked the overarching question of how to find a facial representation that can be readily adapted to several facial analysis tasks and datasets. To this end, we make the following 4 contributions: (a) we introduce, for the first time, a comprehensive evaluation benchmark for facial representation learning consisting of 5 important face analysis tasks. (b) We systematically investigate two ways of large-scale representation learning applied to faces: supervised and unsupervised pre-training. Importantly, we focus our evaluations on the case of few-shot facial learning. (c) We investigate important properties of the training datasets including their size and quality (labelled, unlabelled or even uncurated). (d) To draw our conclusions, we conducted a very large number of experiments. Our main two findings are: (1) Unsupervised pre-training on completely in-the-wild, uncurated data provides consistent and, in some cases, significant accuracy improvements for all facial tasks considered. (2) Many existing facial video datasets seem to have a large amount of redundancy. We will release code, and pre-trained models to facilitate future research.

cs.CV↗

arXiv2022

Tailoring Dynamical Fermionization: Delta kick cooling of a Tonks-Girardeau gas

In one spatial dimension, quantum exchange statistics and interactions are inextricably intertwined. As a manifestation, the expansion dynamics of a Tonks-Girardeau gas is characterized by dynamical fermionization (DF), whereby the momentum distribution approaches that of a spin-polarized Fermi gas. Using a phase-space analysis and the unitary evolution of the one-body reduced density matrix, we show that DF can be tailored and reversed, using a generalization of delta kick cooling (DKC) to interacting systems, establishing a simple protocol to rescale the initial momentum distribution. The protocol applies to both expansions and compressions and can be used for microscopy of quantum correlations.

cond-mat.quant-gas↗

arXiv2022

A gated group sequential design for seamless Phase II/III trial with subpopulation selection

Due to the high cost and high failure rate of Phase III trials, seamless Phase II/III designs are more and more popular to trial efficiency. A potential attraction of Phase II/III design is to allow a randomized proof-of-concept stage prior to committing to the full cost of the Phase III trial. Population selection during the trial allows a trial to adapt and focus investment where it is most likely to provide patient benefit. Motivated by a clinical trial to find the population that potential benefits with dual-primary endpoints progression free survival (PFS) and overall survival (OS), we propose a gated group sequential design for a seamless Phase II/III trial design with population selection. The investigated design controls the familywise error rate and allows multiple interim analyses to enable early stopping for efficacy or futility. Simulations and an illustrative example suggest that the proposed gated group sequential design can have more power than the commonly used classical group sequential design, and reduces the patient's exposure to less effective treatment if the complementary sub-group has less significant treatment effect. The proposed design has the potential to save drug development cost and more quickly fulfill unmet medical needs.

stat.AP↗

arXiv2022

Region-Aware Network: Model Human's Top-Down Visual Perception Mechanism for Crowd Counting

Background noise and scale variation are common problems that have been long recognized in crowd counting. Humans glance at a crowd image and instantly know the approximate number of human and where they are through attention the crowd regions and the congestion degree of crowd regions with a global receptive field. Hence, in this paper, we propose a novel feedback network with Region-Aware block called RANet by modeling humans Top-Down visual perception mechanism. Firstly, we introduce a feedback architecture to generate priority maps that provide prior about candidate crowd regions in input images. The prior enables the RANet pay more attention to crowd regions. Then we design Region-Aware block that could adaptively encode the contextual information into input images through global receptive field. More specifically, we scan the whole input images and its priority maps in the form of column vector to obtain a relevance matrix estimating their similarity. The relevance matrix obtained would be utilized to build global relationships between pixels. Our method outperforms state-of-the-art crowd counting methods on several public datasets.

cs.CV↗