arXiv subjects

Hao Yang

Publications and source records attributed to Hao Yang.

At least 361 records · Page 20Linked to original sources

arXiv2022

General Facial Representation Learning in a Visual-Linguistic Manner

How to learn a universal facial representation that boosts all face analysis tasks? This paper takes one step toward this goal. In this paper, we study the transfer performance of pre-trained models on face analysis tasks and introduce a framework, called FaRL, for general Facial Representation Learning in a visual-linguistic manner. On one hand, the framework involves a contrastive loss to learn high-level semantic meaning from image-text pairs. On the other hand, we propose exploring low-level information simultaneously to further enhance the face representation, by adding a masked image modeling. We perform pre-training on LAION-FACE, a dataset containing large amount of face image-text pairs, and evaluate the representation capability on multiple downstream tasks. We show that FaRL achieves better transfer performance compared with previous pre-trained models. We also verify its superiority in the low-data regime. More importantly, our model surpasses the state-of-the-art methods on face analysis tasks including face parsing and face alignment.

cs.CV↗

arXiv2022

Omni-DETR: Omni-Supervised Object Detection with Transformers

We consider the problem of omni-supervised object detection, which can use unlabeled, fully labeled and weakly labeled annotations, such as image tags, counts, points, etc., for object detection. This is enabled by a unified architecture, Omni-DETR, based on the recent progress on student-teacher framework and end-to-end transformer based object detection. Under this unified architecture, different types of weak labels can be leveraged to generate accurate pseudo labels, by a bipartite matching based filtering mechanism, for the model to learn. In the experiments, Omni-DETR has achieved state-of-the-art results on multiple datasets and settings. And we have found that weak annotations can help to improve detection performance and a mixture of them can achieve a better trade-off between annotation cost and accuracy than the standard complete annotation. These findings could encourage larger object detection datasets with mixture annotations. The code is available at https://github.com/amazon-research/omni-detr.

cs.CV↗

arXiv2022

NLO QCD corrections to pseudoscalar quarkonium production with two heavy flavors in photon-photon collision

We calculate the next-to-leading order (NLO) quantum chromodynamics (QCD) corrections to $γ+γ\to η_c+c+\bar{c}$, $γ+γ\to η_b+b+\bar{b}$ and $γ+γ\to B_c+b+\bar{c}$ processes in the framework of non-relativistic QCD (NRQCD) factorization formalism. The cross sections at the SuperKEKB electron-positron collider, as well as the future collider like the Circular Electron Positron Collider (CEPC), are evaluated. Numerical results indicate that the NLO corrections are significant, and the uncertainties in theoretical predictions with NLO corrections are reduced as expected. Due to the high luminosity of the SuperKEKB collider, the $η_c+c+\bar{c}$ production is hopefully observable in the near future.

hep-ph↗

arXiv2022

New criterions on nonexistence of periodic orbits of planar dynamical systems and their applications

Characterizing existence or not of periodic orbit is a classical problem and it has both theoretical importance and many real applications. Here, several new criterions on nonexistence of periodic orbits of the planar dynamical system $\dot x=y,~\dot y=-g(x)-f(x,y)y$ are obtained in this paper, and by examples showing that these criterions are applicable, but the known ones are invalid to them. Based on these criterions, we further characterize the local topological structures of its equilibrium, which also show that one of the classical results by A.F. Andreev [Amer. Math. Soc. Transl. 8 (1958), 183--207] on local topological classification of the degenerate equilibrium is incomplete. Finally, as another application of these results, we classify the global phase portraits of a planar differential system, which comes from the third question in the list of the 33 questions posed by A. Gasull and also from a mechanical oscillator under suitable restriction to its parameters.

math.DS↗

arXiv2022

Sentiment Word Aware Multimodal Refinement for Multimodal Sentiment Analysis with ASR Errors

Multimodal sentiment analysis has attracted increasing attention and lots of models have been proposed. However, the performance of the state-of-the-art models decreases sharply when they are deployed in the real world. We find that the main reason is that real-world applications can only access the text outputs by the automatic speech recognition (ASR) models, which may be with errors because of the limitation of model capacity. Through further analysis of the ASR outputs, we find that in some cases the sentiment words, the key sentiment elements in the textual modality, are recognized as other words, which makes the sentiment of the text change and hurts the performance of multimodal sentiment models directly. To address this problem, we propose the sentiment word aware multimodal refinement model (SWRM), which can dynamically refine the erroneous sentiment words by leveraging multimodal sentiment clues. Specifically, we first use the sentiment word position detection module to obtain the most possible position of the sentiment word in the text and then utilize the multimodal sentiment word refinement module to dynamically refine the sentiment word embeddings. The refined embeddings are taken as the textual inputs of the multimodal feature fusion module to predict the sentiment labels. We conduct extensive experiments on the real-world datasets including MOSI-Speechbrain, MOSI-IBM, and MOSI-iFlytek and the results demonstrate the effectiveness of our model, which surpasses the current state-of-the-art models on three datasets. Furthermore, our approach can be adapted for other multimodal feature fusion models easily. Data and code are available at https://github.com/albertwy/SWRM.

cs.CL↗

arXiv2022

A GNSS Aided Initial Alignment Method for MEMS-IMU Based on Backtracking Algorithm and Backward Filtering

To obtain a high-accuracy position with SINS(Strapdown Inertial Navigation System), initial alignment needs to determine initial attitude rapidly and accurately. High-accuracy grade IMU(Inertial Measurement Uint) can obtain the initial attitude indenpendently, however, the low-accuracy grade gyroscope doesn't adapt to determine the heading angle, hence the initial attitude matrix will not be obtained. If using large misalignment angle model to estiamting heading angle, the convergence time will become much longer. For solving these two problems, a novel alignment algorithm combined backtracking algorithm and reverse navigation updating method with GNSS(Global Navigation Satellite System) aiding is proposed herein. The simulation and land vehicle test were finished to evaluate the alignment accuracy of the proposed algorithm. The horizontal misalignment is less than 2.3 arcmin and the heading misalignment is less than 10.1 arcmin in test. The proposed algorithm is a feasible and practical alignment method for low-cost IMU to obtain initial attitude in short term and large misalignment condition aided by GNSS.

cs.RO↗

arXiv2022

Text Steganalysis with Attentional LSTM-CNN

With the rapid development of Natural Language Processing (NLP) technologies, text steganography methods have been significantly innovated recently, which poses a great threat to cybersecurity. In this paper, we propose a novel attentional LSTM-CNN model to tackle the text steganalysis problem. The proposed method firstly maps words into semantic space for better exploitation of the semantic feature in texts and then utilizes a combination of Convolutional Neural Networks (CNNs) and Long Short-Term Memory (LSTM) recurrent neural networks to capture both local and long-distance contextual information in steganography texts. In addition, we apply attention mechanism to recognize and attend to important clues within suspicious sentences. After merge feature clues from Convolutional Neural Networks (CNNs) and Recurrent Neural Networks (RNNs), we use a softmax layer to categorize the input text as cover or stego. Experiments showed that our model can achieve the state-of-art result in the text steganalysis task.

cs.MM↗

arXiv2022

Remarkably Enhanced Dynamic Oxygen Migration on Graphene Oxide Supported by Copper Substrate

The dynamic covalent properties of graphene oxide (GO) are of fundamental interest to a broad range of scientific areas and technological applications. It remains a challenge to access the feasible dynamic reactions for reversibly breaking/reforming covalent bonds of oxygen functional groups on GO, although these reactions can be induced by photonic or mechanical routes, or mediated by adsorbed water. Here, using the density functional theory calculations, we demonstrate the remarkably enhanced dynamic oxygen migration along the basal plane of GO supported by copper substrate (GO@copper), with the C-O bond breaking reaction and proton transfer between the neighboring epoxy and hydroxyl groups. Compared to that on GO, the energy barrier of oxygen migration on GO@copper is sharply reduced to be less than or comparable to thermal fluctuations, and meanwhile the crystallographic match between GO and copper substrate induces new oxygen migration paths on GO@copper. This work sheds light on the understanding of metal substrate-enhanced dynamic properties of GO, and evidences the strategy to tune the activity of two-dimension-interfacial oxygen groups for various potential applications.

cond-mat.mtrl-sci↗

arXiv2022

Forecast-based Multi-aspect Framework for Multivariate Time-series Anomaly Detection

Today's cyber-world is vastly multivariate. Metrics collected at extreme varieties demand multivariate algorithms to properly detect anomalies. However, forecast-based algorithms, as widely proven approaches, often perform sub-optimally or inconsistently across datasets. A key common issue is they strive to be one-size-fits-all but anomalies are distinctive in nature. We propose a method that tailors to such distinction. Presenting FMUAD - a Forecast-based, Multi-aspect, Unsupervised Anomaly Detection framework. FMUAD explicitly and separately captures the signature traits of anomaly types - spatial change, temporal change and correlation change - with independent modules. The modules then jointly learn an optimal feature representation, which is highly flexible and intuitive, unlike most other models in the category. Extensive experiments show our FMUAD framework consistently outperforms other state-of-the-art forecast-based anomaly detectors.

cs.LG↗

arXiv2022

Rethinking Feature Uncertainty in Stochastic Neural Networks for Adversarial Robustness

It is well-known that deep neural networks (DNNs) have shown remarkable success in many fields. However, when adding an imperceptible magnitude perturbation on the model input, the model performance might get rapid decrease. To address this issue, a randomness technique has been proposed recently, named Stochastic Neural Networks (SNNs). Specifically, SNNs inject randomness into the model to defend against unseen attacks and improve the adversarial robustness. However, existed studies on SNNs mainly focus on injecting fixed or learnable noises to model weights/activations. In this paper, we find that the existed SNNs performances are largely bottlenecked by the feature representation ability. Surprisingly, simply maximizing the variance per dimension of the feature distribution leads to a considerable boost beyond all previous methods, which we named maximize feature distribution variance stochastic neural network (MFDV-SNN). Extensive experiments on well-known white- and black-box attacks show that MFDV-SNN achieves a significant improvement over existing methods, which indicates that it is a simple but effective method to improve model robustness.

cs.LG↗

arXiv2021

Diformer: Directional Transformer for Neural Machine Translation

Autoregressive (AR) and Non-autoregressive (NAR) models have their own superiority on the performance and latency, combining them into one model may take advantage of both. Current combination frameworks focus more on the integration of multiple decoding paradigms with a unified generative model, e.g. Masked Language Model. However, the generalization can be harmful to the performance due to the gap between training objective and inference. In this paper, we aim to close the gap by preserving the original objective of AR and NAR under a unified framework. Specifically, we propose the Directional Transformer (Diformer) by jointly modelling AR and NAR into three generation directions (left-to-right, right-to-left and straight) with a newly introduced direction variable, which works by controlling the prediction of each token to have specific dependencies under that direction. The unification achieved by direction successfully preserves the original dependency assumption used in AR and NAR, retaining both generalization and performance. Experiments on 4 WMT benchmarks demonstrate that Diformer outperforms current united-modelling works with more than 1.5 BLEU points for both AR and NAR decoding, and is also competitive to the state-of-the-art independent AR and NAR models.

cs.CL↗

arXiv2021

Self-Distillation Mixup Training for Non-autoregressive Neural Machine Translation

Recently, non-autoregressive (NAT) models predict outputs in parallel, achieving substantial improvements in generation speed compared to autoregressive (AT) models. While performing worse on raw data, most NAT models are trained as student models on distilled data generated by AT teacher models, which is known as sequence-level Knowledge Distillation. An effective training strategy to improve the performance of AT models is Self-Distillation Mixup (SDM) Training, which pre-trains a model on raw data, generates distilled data by the pre-trained model itself and finally re-trains a model on the combination of raw data and distilled data. In this work, we aim to view SDM for NAT models, but find directly adopting SDM to NAT models gains no improvements in terms of translation quality. Through careful analysis, we observe the invalidation is correlated to Modeling Diversity and Confirmation Bias between the AT teacher model and the NAT student models. Based on these findings, we propose an enhanced strategy named SDMRT by adding two stages to classic SDM: one is Pre-Rerank on self-distilled data, the other is Fine-Tune on Filtered teacher-distilled data. Our results outperform baselines by 0.6 to 1.2 BLEU on multiple NAT models. As another bonus, for Iterative Refinement NAT models, our methods can outperform baselines within half iteration number, which means 2X acceleration.

cs.CL↗

arXiv2021

Joint-training on Symbiosis Networks for Deep Nueral Machine Translation models

Deep encoders have been proven to be effective in improving neural machine translation (NMT) systems, but it reaches the upper bound of translation quality when the number of encoder layers exceeds 18. Worse still, deeper networks consume a lot of memory, making it impossible to train efficiently. In this paper, we present Symbiosis Networks, which include a full network as the Symbiosis Main Network (M-Net) and another shared sub-network with the same structure but less layers as the Symbiotic Sub Network (S-Net). We adopt Symbiosis Networks on Transformer-deep (m-n) architecture and define a particular regularization loss $\mathcal{L}_τ$ between the M-Net and S-Net in NMT. We apply joint-training on the Symbiosis Networks and aim to improve the M-Net performance. Our proposed training strategy improves Transformer-deep (12-6) by 0.61, 0.49 and 0.69 BLEU over the baselines under classic training on WMT'14 EN->DE, DE->EN and EN->FR tasks. Furthermore, our Transformer-deep (12-6) even outperforms classic Transformer-deep (18-6).

cs.CL↗

arXiv2021

Delayed Propagation Transformer: A Universal Computation Engine towards Practical Control in Cyber-Physical Systems

Multi-agent control is a central theme in the Cyber-Physical Systems (CPS). However, current control methods either receive non-Markovian states due to insufficient sensing and decentralized design, or suffer from poor convergence. This paper presents the Delayed Propagation Transformer (DePT), a new transformer-based model that specializes in the global modeling of CPS while taking into account the immutable constraints from the physical world. DePT induces a cone-shaped spatial-temporal attention prior, which injects the information propagation and aggregation principles and enables a global view. With physical constraint inductive bias baked into its design, our DePT is ready to plug and play for a broad class of multi-agent systems. The experimental results on one of the most challenging CPS -- network-scale traffic signal control system in the open world -- show that our model outperformed the state-of-the-art expert methods on synthetic and real-world datasets. Our codes are released at: https://github.com/VITA-Group/DePT.

cs.AI↗

arXiv2021

Large deviation principles for SDEs under locally weak monotonicity conditions

This paper establishes a Freidlin-Wentzell large deviation principle for stochastic differential equations(SDEs) under locally weak monotonicity conditions and Lyapunov conditions. We illustrate the main result of the paper by showing that it can be applied to SDEs with non-Lipschitzian coefficients, which can not be covered in the existing literature. These include the interesting biological models like stochastic Duffing-van der Pol oscillator model, stochastic SIR model, etc.

math.PR↗

arXiv2021

How Does Adversarial Fine-Tuning Benefit BERT?

Adversarial training (AT) is one of the most reliable methods for defending against adversarial attacks in machine learning. Variants of this method have been used as regularization mechanisms to achieve SOTA results on NLP benchmarks, and they have been found to be useful for transfer learning and continual learning. We search for the reasons for the effectiveness of AT by contrasting vanilla and adversarially fine-tuned BERT models. We identify partial preservation of BERT's syntactic abilities during fine-tuning as the key to the success of AT. We observe that adversarially fine-tuned models remain more faithful to BERT's language modeling behavior and are more sensitive to the word order. As concrete examples of syntactic abilities, an adversarially fine-tuned model could have an advantage of up to 38% on anaphora agreement and up to 11% on dependency parsing. Our analysis demonstrates that vanilla fine-tuning oversimplifies the sentence representation by focusing heavily on a small subset of words. AT, however, moderates the effect of these influential words and encourages representational diversity. This allows for a more hierarchical representation of a sentence and leads to the mitigation of BERT's loss of syntactic abilities.

cs.CL↗

arXiv2021

Quasinormal Modes and Thermodynamics of Regular Black Holes

By applying the dimensionless scheme, we investigate the quasinormal modes and phase transitions analytically for three types of regular black holes. The universal deviations to the first law of mechanics in regular black holes are proved. Meanwhile, we verify that second order phase transitions and Davies points still exist in these three models. In addition, we calculate their quasinormal modes in the eikonal limit by applying the light ring/quasinormal mode correspondence, and discuss the spiral-like shapes and the relations between the quasinormal modes and phase transitions. As the main result, we show that spiral-like shapes in the complex frequency plane are closely related to the parameterization, namely in some particular units the spiral-like shapes will emerge in the models, which may not be of the spiral behaviors reported by other authors. We also discover a universal property of regular black holes, i.e., the imaginary parts of their QNMs do not vanish for the extreme cases, which does not appear in singular black holes, such as the Reissner-Nordström and Kerr black holes, etc.

gr-qc↗

arXiv2021

Initial energy deposition and initiation mechanism of nanosecond laser damage caused by KDP surface micro-defects

To enable an exploration of the initiation mechanism of nanosecond laser damage on a potassium dihydrogen phosphate (KDP) surface, a defect-assisted energy deposition model is developed that involves light intensity enhancement and a sub-band gap energy level structure. The simulations provide an explanation on why the laser-induced damage threshold (LIDT) of the KDP crystal is two orders of magnitude lower than the theoretical value. The model is verified by use of the transient images that appear during the laser damage. In addition, the dimensions of the "dangerous" surface defects that are the most sensitive to the laser damage are proposed. This work enables clarification on the initial energy deposition (IED) and initiation mechanism of the nanosecond laser damage caused by the KDP surface defects on micro-nano scale. It is helpful in understanding the laser-matter interactions and to improve the processing technique for high quality optical components.

physics.app-ph↗