arXiv subjects

Tao Wang

Publications and source records attributed to Tao Wang.

At least 541 records · Page 30Linked to original sources

arXiv2023

Robust Remote Sensing Scene Classification with Multi-View Voting and Entropy Ranking

Deep convolutional neural networks have been widely used in scene classification of remotely sensed images. In this work, we propose a robust learning method for the task that is secure against partially incorrect categorization of images. Specifically, we remove and correct errors in the labels progressively by iterative multi-view voting and entropy ranking. At each time step, we first divide the training data into disjoint parts for separate training and voting. The unanimity in the voting reveals the correctness of the labels, so that we can train a strong model with only the images with unanimous votes. In addition, we adopt entropy as an effective measure for prediction uncertainty, in order to partially recover labeling errors by ranking and selection. We empirically demonstrate the superiority of the proposed method on the WHU-RS19 dataset and the AID dataset.

cs.CV↗

arXiv2023

UnifySpeech: A Unified Framework for Zero-shot Text-to-Speech and Voice Conversion

Text-to-speech (TTS) and voice conversion (VC) are two different tasks both aiming at generating high quality speaking voice according to different input modality. Due to their similarity, this paper proposes UnifySpeech, which brings TTS and VC into a unified framework for the first time. The model is based on the assumption that speech can be decoupled into three independent components: content information, speaker information, prosody information. Both TTS and VC can be regarded as mining these three parts of information from the input and completing the reconstruction of speech. For TTS, the speech content information is derived from the text, while in VC it's derived from the source speech, so all the remaining units are shared except for the speech content extraction module in the two tasks. We applied vector quantization and domain constrain to bridge the gap between the content domains of TTS and VC. Objective and subjective evaluation shows that by combining the two task, TTS obtains better speaker modeling ability while VC gets hold of impressive speech content decoupling capability.

cs.SD↗

arXiv2022

Superconductivity in the Uniform Electron Gas: Irrelevance of Kohn-Luttinger Mechanism

We study the Cooper instability in jellium model in the controlled regime of small to intermediate values of the Coulomb parameter $r_s \leq 2$. We confirm that superconductivity naturally emerges from purely repulsive interactions described by the Kukkonen-Overhauser vertex function. By employing the implicit renormalization approach we reveal that even in the small-$r_s$ limit, the dominant mechanism behind Cooper instability is based on dynamic screening of the Coulomb interaction--accurately captured by the random phase approximation, whereas the Kohn-Luttinger contribution is negligibly small and, thus, not relevant.

cond-mat.supr-con↗

arXiv2022

SODAR: Segmenting Objects by DynamicallyAggregating Neighboring Mask Representations

Recent state-of-the-art one-stage instance segmentation model SOLO divides the input image into a grid and directly predicts per grid cell object masks with fully-convolutional networks, yielding comparably good performance as traditional two-stage Mask R-CNN yet enjoying much simpler architecture and higher efficiency. We observe SOLO generates similar masks for an object at nearby grid cells, and these neighboring predictions can complement each other as some may better segment certain object part, most of which are however directly discarded by non-maximum-suppression. Motivated by the observed gap, we develop a novel learning-based aggregation method that improves upon SOLO by leveraging the rich neighboring information while maintaining the architectural efficiency. The resulting model is named SODAR. Unlike the original per grid cell object masks, SODAR is implicitly supervised to learn mask representations that encode geometric structure of nearby objects and complement adjacent representations with context. The aggregation method further includes two novel designs: 1) a mask interpolation mechanism that enables the model to generate much fewer mask representations by sharing neighboring representations among nearby grid cells, and thus saves computation and memory; 2) a deformable neighbour sampling mechanism that allows the model to adaptively adjust neighbor sampling locations thus gathering mask representations with more relevant context and achieving higher performance. SODAR significantly improves the instance segmentation performance, e.g., it outperforms a SOLO model with ResNet-101 backbone by 2.2 AP on COCO \texttt{test} set, with only about 3\% additional computation. We further show consistent performance gain with the SOLOv2 model.

cs.CV↗

arXiv2022

Ultra-High-Definition Low-Light Image Enhancement: A Benchmark and Transformer-Based Method

As the quality of optical sensors improves, there is a need for processing large-scale images. In particular, the ability of devices to capture ultra-high definition (UHD) images and video places new demands on the image processing pipeline. In this paper, we consider the task of low-light image enhancement (LLIE) and introduce a large-scale database consisting of images at 4K and 8K resolution. We conduct systematic benchmarking studies and provide a comparison of current LLIE algorithms. As a second contribution, we introduce LLFormer, a transformer-based low-light enhancement method. The core components of LLFormer are the axis-based multi-head self-attention and cross-layer attention fusion block, which significantly reduces the linear complexity. Extensive experiments on the new dataset and existing public datasets show that LLFormer outperforms state-of-the-art methods. We also show that employing existing LLIE methods trained on our benchmark as a pre-processing step significantly improves the performance of downstream tasks, e.g., face detection in low-light conditions. The source code and pre-trained models are available at https://github.com/TaoWangzj/LLFormer.

cs.CV↗

arXiv2022

Emotion Selectable End-to-End Text-based Speech Editing

Text-based speech editing allows users to edit speech by intuitively cutting, copying, and pasting text to speed up the process of editing speech. In the previous work, CampNet (context-aware mask prediction network) is proposed to realize text-based speech editing, significantly improving the quality of edited speech. This paper aims at a new task: adding emotional effect to the editing speech during the text-based speech editing to make the generated speech more expressive. To achieve this task, we propose Emo-CampNet (emotion CampNet), which can provide the option of emotional attributes for the generated speech in text-based speech editing and has the one-shot ability to edit unseen speakers' speech. Firstly, we propose an end-to-end emotion-selectable text-based speech editing model. The key idea of the model is to control the emotion of generated speech by introducing additional emotion attributes based on the context-aware mask prediction network. Secondly, to prevent the emotion of the generated speech from being interfered by the emotional components in the original speech, a neutral content generator is proposed to remove the emotion from the original speech, which is optimized by the generative adversarial framework. Thirdly, two data augmentation methods are proposed to enrich the emotional and pronunciation information in the training set, which can enable the model to edit the unseen speaker's speech. The experimental results that 1) Emo-CampNet can effectively control the emotion of the generated speech in the process of text-based speech editing; And can edit unseen speakers' speech. 2) Detailed ablation experiments further prove the effectiveness of emotional selectivity and data augmentation methods. The demo page is available at https://hairuo55.github.io/Emo-CampNet/

cs.SD↗

arXiv2022

High temperature AdS black holes are low temperature quantum phonon gases

We report a precise match between the high temperature $(D+2)$-dimensional Tangherlini-AdS black hole and the low temperature quantum phonon gas in $D$-dimensional nonmetallic crystals residing in $(D+1)$-dimensional flat spacetime. The match is realized by use of the recently proposed restricted phase space formalism for black hole thermodynamics, and the result can be viewed as a novel contribution to the AdS/CMT correspondence on a quantitative level.

hep-th↗

arXiv2022

Improving Named Entity Recognition by External Context Retrieving and Cooperative Learning

Recent advances in Named Entity Recognition (NER) show that document-level contexts can significantly improve model performance. In many application scenarios, however, such contexts are not available. In this paper, we propose to find external contexts of a sentence by retrieving and selecting a set of semantically relevant texts through a search engine, with the original sentence as the query. We find empirically that the contextual representations computed on the retrieval-based input view, constructed through the concatenation of a sentence and its external contexts, can achieve significantly improved performance compared to the original input view based only on the sentence. Furthermore, we can improve the model performance of both input views by Cooperative Learning, a training method that encourages the two input views to produce similar contextual representations or output label distributions. Experiments show that our approach can achieve new state-of-the-art performance on 8 NER data sets across 5 domains.

cs.CL↗

arXiv2022

Interplay between spin wave and magnetic vortex

In this paper, the interplay between spin wave and magnetic vortex is studied. We find three types of magnon scatterings: skew scattering, symmetric side deflection and back reflection, which associate with respectively magnetic topology, energy density distribution and linear momentum transfer torque within vortex. The vortex core exhibits two translational modes: the intrinsic circular mode and a coercive elliptical mode, which can be excited based on permanent and periodic magnon spin-transfer torque effects of spin wave. Lastly, we propose a vortex-based spin wave valve in which via inhomogeneity modulation we access arbitrary control of the phase shift.

cond-mat.mes-hall↗

arXiv2022

Peering into the Milky Way by FAST: I. Exquisite HI structures in the inner Galactic disk from the piggyback line observations of the FAST GPPS survey

Neutral hydrogen (HI) is the fundamental component of the interstellar medium. The Galactic Plane Pulsar Snapshot (GPPS) survey is designed for hunting pulsars by using the Five-hundred-meter Aperture Spherical radio Telescope (FAST) from the visible Galactic plane within $|b| \leq 10^{\circ}$. The survey observations are conducted with the L-band 19-beam receiver in the frequency range of 1.0 $-$ 1.5 GHz, and each pointing has an integration time of 5 minutes. The piggyback spectral data simultaneously recorded during the FAST GPPS survey are great resources for studies on the Galactic HI distribution and ionized gas. We process the piggyback HI data of the FAST GPPS survey in the region of $33^{\circ} \leq l \leq 55^{\circ}$ and $|b| \leq 2^{\circ}$. The rms of the data cube is found to be approximately 40 mK at a velocity resolution of $0.1$ km s$^{-1}$, placing it the most sensitive observations of the Galactic HI by far. The high velocity resolution and high sensitivity of the FAST GPPS HI data enable us to detect weak exquisite HI structures in the interstellar medium. HI absorption line with great details can be obtained against bright continuum sources. The FAST GPPS survey piggyback HI data cube will be released and updated on the web: http://zmtt.bao.ac.cn/MilkyWayFAST/.

astro-ph.GA↗

arXiv2022

High-Level Synthesis for Packet-Processing Pipelines

Compiling high-level programs to target high-speed packet-processing pipelines is a challenging combinatorial optimization problem. The compiler must configure the pipeline's resources to match the high-level semantics of the program, while packing all of the program's computation into the pipeline's limited resources. State of the art approaches tackle individual aspects of this problem. Yet, they miss opportunities to efficiently produce globally high-quality outcomes. We argue that High-Level Synthesis (HLS), previously applied to ASIC/FPGA design, is the right framework to decompose the compilation problem for pipelines into smaller pieces with modular solutions. We design an HLS-based compiler that works in three phases. Transformation rewrites programs to use more abundant pipeline resources, avoiding scarce ones. Synthesis breaks complex transactional code into configurations of pipelined compute units. Allocation maps the program's compute and memory to the hardware resources. We prototype these ideas in a compiler, CaT, which targets the Tofino pipeline and a cycle-accurate simulator of a Verilog hardware model of an RMT pipeline. CaT can handle programs that existing compilers cannot currently run on pipelines, generating code faster than existing compilers, while using fewer pipeline resources.

cs.NI↗

arXiv2022

ALMA Lensing Cluster Survey: $HST$ and $Spitzer$ Photometry of 33 Lensed Fields Built with CHArGE

We present a set of multi-wavelength mosaics and photometric catalogs in the ALMA lensing cluster survey (ALCS) fields. The catalogs were built by reprocessing of archival data from the CHArGE compilation, taken by the $\textit{Hubble Space Telescope}$ ($\textit{HST}$) in the RELICS, CLASH and Hubble Frontier Fields. Additionally we have reconstructed the $\textit{Spitzer}$ IRAC 3.6 and 4.5 $μ$m mosaics, by utilising all the available archival IRSA/SHA exposures. To alleviate the effect of blending in such a crowded region, we have modelled the $\textit{Spitzer}$ photometry by convolving the $\textit{HST}$ detection image with the $\textit{Spitzer}$ PSF using the novel $\texttt{golfir}$ software. The final catalogs contain 218,000 sources, covering a combined area of 690 arcmin$^2$. These catalogs will serve as an important tool in aiding the search of the sub-mm galaxies in future ALMA surveys, as well as follow ups of the $\textit{HST}$ dark - IRAC sources. Coupled with the available $\textit{HST}$ photometry the addition of the 3.6 and 4.5 $μ$m bands will allow us to place a better constraint on photometric redshifts and stellar masses of these objects, thus giving us an opportunity to identify high-redshift candidates for spectroscopic follow ups and answer the important questions regarding the epoch of reionization and formation of first galaxies.

astro-ph.GA↗

arXiv2022

The Volumetric Extended-Schmidt Law: A Unity Slope

We investigate the extended-Schmidt (ES) law in volume densities ($ρ_{\rm SFR}$ $\propto$ $(ρ_{\rm gas}ρ_{\rm star}^{0.5})^{α^{\rm VES}}$) for spatially-resolved regions in spiral, dwarf, and ultra-diffuse galaxies (UDGs), and compare to the volumetric Kennicutt-Schmidt (KS) law ($ρ_{\rm SFR}$ $\propto$ $ρ_{\rm gas}^{α^{\rm VKS}}$). We first characterize these star formation laws in individual galaxies using a sample of 11 spirals, finding median slopes $α^{\rm VES}$=0.98 and $α^{\rm VKS}$=1.42, with a galaxy-to-galaxy rms fluctuation that is substantially smaller for the volumetric ES law (0.18 vs 0.41). By combining all regions in spirals with those in additional 13 dwarfs and one UDG into one single dataset, it is found that the rms scatter of the volumetric ES law at given x-axis is 0.25 dex, also smaller than that of the volumetric KS law (0.34 dex). At the extremely low gas density regime as offered by the UDG, the volumetric KS law breaks down but the volumetric ES law still holds. On the other hand, as compared to the surface density ES law, the volumetric ES law instead has a slightly larger rms scatter, consistent with the scenario that the ES law has an intrinsic slope of $α^{\rm VES} \equiv$1 but the additional observational error of the scale height increases the uncertainty of the volume density. The unity slope of the ES law implies that the star formation efficiency (=$ρ_{\rm SFR}$/$ρ_{\rm gas}$) is regulated by the quantity that is related to the $ρ_{\rm star}^{0.5}$.

astro-ph.GA↗

arXiv2022

Thermally Induced Structural Evolution and Nanoscale Interfacial Dynamics in Layered Metal Chalcogenides

Layered chalcogenides including Bi2Te3, Sb2Te3, Bi-Sb-Te ternary alloys and heterostructures are known as great thermoelectric, topological insulators and recently highlighted as plasmonic building blocks beyond noble metals. Here, we conduct a joint in situ transmission electron microscopy (in situ TEM) and density functional theory (DFT) calculations to investigate the temperature dependent nanoscale dynamics, interfacial properties and further identifying the role of native defects and edge configurations in anisotropic sublimations of Bi2Te3-Sb2Te3 in-plane heterostructure and Sb2-xBixTe3 alloy. Structural dynamics including edge evolution, formation, expansion, and coalescence of thermally induced polygonal nanopores are reported. The nanopores appear to be initiated by preferential dissociation of chalcogenide species (Te) from the center, heterointerface and edges in the heterostructure and only from the outer edges in the alloy counterpart. This results in a reduced thermal stability and significantly different sublimation pathways of the heterostructure. Furthermore, triangular and quasi hexagonal configurations are observed to be the dominant nanopores configurations in the heterostructure. Additionally, our DFT calculations provide a mechanistic understanding on the role of native defects and edge formation energies, revealing the antisite defects TeBi to be the dominant native defect in a Te-rich condition and playing a key role on the defects assisted sublimation. These findings significantly impact our understanding of controlling the nanoscale sublimation dynamics and can ultimately assist us in designing tunable low-dimensional chalcogenides.

cond-mat.mtrl-sci↗

arXiv2022

Towards Real World HDRTV Reconstruction: A Data Synthesis-based Approach

Existing deep learning based HDRTV reconstruction methods assume one kind of tone mapping operators (TMOs) as the degradation procedure to synthesize SDRTV-HDRTV pairs for supervised training. In this paper, we argue that, although traditional TMOs exploit efficient dynamic range compression priors, they have several drawbacks on modeling the realistic degradation: information over-preservation, color bias and possible artifacts, making the trained reconstruction networks hard to generalize well to real-world cases. To solve this problem, we propose a learning-based data synthesis approach to learn the properties of real-world SDRTVs by integrating several tone mapping priors into both network structures and loss functions. In specific, we design a conditioned two-stream network with prior tone mapping results as a guidance to synthesize SDRTVs by both global and local transformations. To train the data synthesis network, we form a novel self-supervised content loss to constraint different aspects of the synthesized SDRTVs at regions with different brightness distributions and an adversarial loss to emphasize the details to be more realistic. To validate the effectiveness of our approach, we synthesize SDRTV-HDRTV pairs with our method and use them to train several HDRTV reconstruction networks. Then we collect two inference datasets containing both labeled and unlabeled real-world SDRTVs, respectively. Experimental results demonstrate that, the networks trained with our synthesized data generalize significantly better to these two real-world datasets than existing solutions.

cs.CV↗

arXiv2022

SJ-HD^2R: Selective Joint High Dynamic Range and Denoising Imaging for Dynamic Scenes

Ghosting artifacts, motion blur, and low fidelity in highlight are the main challenges in High Dynamic Range (HDR) imaging from multiple Low Dynamic Range (LDR) images. These issues come from using the medium-exposed image as the reference frame in previous methods. To deal with them, we propose to use the under-exposed image as the reference to avoid these issues. However, the heavy noise in dark regions of the under-exposed image becomes a new problem. Therefore, we propose a joint HDR and denoising pipeline, containing two sub-networks: (i) a pre-denoising network (PreDNNet) to adaptively denoise input LDRs by exploiting exposure priors; (ii) a pyramid cascading fusion network (PCFNet), introducing an attention mechanism and cascading structure in a multi-scale manner. To further leverage these two paradigms, we propose a selective and joint HDR and denoising (SJ-HD$^2$R) imaging framework, utilizing scenario-specific priors to conduct the path selection with an accuracy of more than 93.3$\%$. We create the first joint HDR and denoising benchmark dataset, which contains a variety of challenging HDR and denoising scenes and supports the switching of the reference image. Extensive experiment results show that our method achieves superior performance to previous methods.

eess.IV↗

arXiv2022

Atomically localized ingredient-dependent interface phonon in heterogeneous solids

Phonons are the primary heat carriers in non-metallic solids. In compositionally heterogeneous materials, the thermal properties are believed to be mainly governed by the disrupted phonon transport due to mass disorder and strain fluctuations, while the effects of compositional fluctuation induced local phonon states are usually ignored. Here, by scanning transmission electron microscopy electron energy loss spectroscopy and sophisticated calculations, we identify the vibrational properties of ingredient-dependent interface phonon modes in AlxGa1-xN and quantify their various contributions to the local interface thermal conductance. We demonstrate that atomic-scale compositional fluctuation has significant influence on the vibrational thermodynamic properties, highly affecting the mode ratio and vibrational amplitude of interface phonon modes and subsequently redistributing their modal contribution to the ITC. Our work provides fundamental insights into understanding of local phonon-boundary interactions in nanoscale inhomogeneities, which reveal new opportunities for optimization of thermal properties via engineering ingredient distribution.

cond-mat.mtrl-sci↗

arXiv2022

Text Enhancement for Paragraph Processing in End-to-End Code-switching TTS

Current end-to-end code-switching Text-to-Speech (TTS) can already generate high quality two languages speech in the same utterance with single speaker bilingual corpora. When the speakers of the bilingual corpora are different, the naturalness and consistency of the code-switching TTS will be poor. The cross-lingual embedding layers structure we proposed makes similar syllables in different languages relevant, thus improving the naturalness and consistency of generated speech. In the end-to-end code-switching TTS, there exists problem of prosody instability when synthesizing paragraph text. The text enhancement method we proposed makes the input contain prosodic information and sentence-level context information, thus improving the prosody stability of paragraph text. Experimental results demonstrate the effectiveness of the proposed methods in the naturalness, consistency, and prosody stability. In addition to Mandarin and English, we also apply these methods to Shanghaiese and Cantonese corpora, proving that the methods we proposed can be extended to other languages to build end-to-end code-switching TTS system.

cs.SD↗