arXiv subjects

Zhuo Chen

Publications and source records attributed to Zhuo Chen.

At least 307 records · Page 17Linked to original sources

arXiv2023

HyperStyle3D: Text-Guided 3D Portrait Stylization via Hypernetworks

Portrait stylization is a long-standing task enabling extensive applications. Although 2D-based methods have made great progress in recent years, real-world applications such as metaverse and games often demand 3D content. On the other hand, the requirement of 3D data, which is costly to acquire, significantly impedes the development of 3D portrait stylization methods. In this paper, inspired by the success of 3D-aware GANs that bridge 2D and 3D domains with 3D fields as the intermediate representation for rendering 2D images, we propose a novel method, dubbed HyperStyle3D, based on 3D-aware GANs for 3D portrait stylization. At the core of our method is a hyper-network learned to manipulate the parameters of the generator in a single forward pass. It not only offers a strong capacity to handle multiple styles with a single model, but also enables flexible fine-grained stylization that affects only texture, shape, or local part of the portrait. While the use of 3D-aware GANs bypasses the requirement of 3D data, we further alleviate the necessity of style images with the CLIP model being the stylization guidance. We conduct an extensive set of experiments across the style, attribute, and shape, and meanwhile, measure the 3D consistency. These experiments demonstrate the superior capability of our HyperStyle3D model in rendering 3D-consistent images in diverse styles, deforming the face shape, and editing various attributes.

cs.CV↗

arXiv2023

Chandra X-ray Measurement of Gas-phase Heavy Element Abundances in the Central Parsec of the Galaxy

Elemental abundances are key to our understanding of star formation and evolution in the Galactic center. Previous work on this topic has been based on infrared (IR) observations, but X-ray observations have the potential of constraining the abundance of heavy elements, mainly through their K-shell emission lines. Using 5.7 Ms Chandra observations, we provide the first abundance measurement of Si, S, Ar, Ca and Fe, in four prominent diffuse X-ray features located in the central parsec of the Galaxy, which are the manifestation of shock-heated hot gas. A two-temperature, non-equilibrium ionization spectral model is employed to derive the abundances of these five elements. In this procedure, a degeneracy is introduced due to uncertainties in the composition of light elements, in particular, H, C and N. Assuming that the hot gas is H-depleted but C- and N-enriched, as would be expected for a standard scenario in which the hot gas is dominated by Wolf-Rayet star winds, the spectral fit finds a generally subsolar abundance for the heavy elements. If, instead, the light elements had a solar-like abundance, the heavy elements have a fitted abundance of $\sim$1--2 solar. The $α$/Fe abundance ratio, on the other hand, is mostly supersolar and insensitive to the exact composition of the light elements. These results are robust against potential biases due to either a moderate spectral S/N or the presence of non-thermal components. Implications of the measured abundances for the Galactic center environment are addressed.

astro-ph.GA↗

arXiv2023

Evidence of a Decreased Binary Fraction for Massive Stars Within 20 Milliparsecs of the Supermassive Black Hole at the Galactic Center

We present the results of the first systematic search for spectroscopic binaries within the central 2 x 3 arcsec$^2$ around the supermassive black hole at the center of the Milky Way galaxy. This survey is based primarily on over a decade of adaptive optics-fed integral-field spectroscopy (R$\sim$4000), obtained as part of the Galactic Center Orbits Initiative at Keck Observatory, and has a limiting $K$'-band magnitude of 15.8, which is at least 4 magnitudes deeper than previous spectroscopic searches for binaries at larger radii within the central nuclear star cluster. From this primary dataset, over 600 new radial velocities are extracted and reported, increasing by a factor of 3 the number of such measurements. We find no significant periodic signals in our sample of 28 stars, of which 16 are massive, young (main-sequence B) stars and 12 are low-mass, old (M and K giant) stars. Using Monte Carlo simulations, we derive upper limits on the intrinsic binary star fraction for the young star population at 47% (at 95% confidence) located $\sim$20 mpc from the black hole. The young star binary fraction is significantly lower than that observed in the field (70%). This result is consistent with a scenario in which the central supermassive black hole drives nearby stellar binaries to merge or be disrupted and may have important implications for the production of gravitational waves and hypervelocity stars.

astro-ph.GA↗

arXiv2023

Target Sound Extraction with Variable Cross-modality Clues

Automatic target sound extraction (TSE) is a machine learning approach to mimic the human auditory perception capability of attending to a sound source of interest from a mixture of sources. It often uses a model conditioned on a fixed form of target sound clues, such as a sound class label, which limits the ways in which users can interact with the model to specify the target sounds. To leverage variable number of clues cross modalities available in the inference phase, including a video, a sound event class, and a text caption, we propose a unified transformer-based TSE model architecture, where a multi-clue attention module integrates all the clues across the modalities. Since there is no off-the-shelf benchmark to evaluate our proposed approach, we build a dataset based on public corpora, Audioset and AudioCaps. Experimental results for seen and unseen target-sound evaluation sets show that our proposed TSE model can effectively deal with a varying number of clues which improves the TSE performance and robustness against partially compromised clues.

eess.AS↗

arXiv2023

Speak Foreign Languages with Your Own Voice: Cross-Lingual Neural Codec Language Modeling

We propose a cross-lingual neural codec language model, VALL-E X, for cross-lingual speech synthesis. Specifically, we extend VALL-E and train a multi-lingual conditional codec language model to predict the acoustic token sequences of the target language speech by using both the source language speech and the target language text as prompts. VALL-E X inherits strong in-context learning capabilities and can be applied for zero-shot cross-lingual text-to-speech synthesis and zero-shot speech-to-speech translation tasks. Experimental results show that it can generate high-quality speech in the target language via just one speech utterance in the source language as a prompt while preserving the unseen speaker's voice, emotion, and acoustic environment. Moreover, VALL-E X effectively alleviates the foreign accent problems, which can be controlled by a language ID. Audio samples are available at \url{https://aka.ms/vallex}.

cs.CL↗

arXiv2023

View Consistency Aware Holistic Triangulation for 3D Human Pose Estimation

The rapid development of multi-view 3D human pose estimation (HPE) is attributed to the maturation of monocular 2D HPE and the geometry of 3D reconstruction. However, 2D detection outliers in occluded views due to neglect of view consistency, and 3D implausible poses due to lack of pose coherence, remain challenges. To solve this, we introduce a Multi-View Fusion module to refine 2D results by establishing view correlations. Then, Holistic Triangulation is proposed to infer the whole pose as an entirety, and anatomy prior is injected to maintain the pose coherence and improve the plausibility. Anatomy prior is extracted by PCA whose input is skeletal structure features, which can factor out global context and joint-by-joint relationship from abstract to concrete. Benefiting from the closed-form solution, the whole framework is trained end-to-end. Our method outperforms the state of the art in both precision and plausibility which is assessed by a new metric.

cs.CV↗

arXiv2023

Tele-Knowledge Pre-training for Fault Analysis

In this work, we share our experience on tele-knowledge pre-training for fault analysis, a crucial task in telecommunication applications that requires a wide range of knowledge normally found in both machine log data and product documents. To organize this knowledge from experts uniformly, we propose to create a Tele-KG (tele-knowledge graph). Using this valuable data, we further propose a tele-domain language pre-training model TeleBERT and its knowledge-enhanced version, a tele-knowledge re-training model KTeleBERT. which includes effective prompt hints, adaptive numerical data encoding, and two knowledge injection paradigms. Concretely, our proposal includes two stages: first, pre-training TeleBERT on 20 million tele-related corpora, and then re-training it on 1 million causal and machine-related corpora to obtain KTeleBERT. Our evaluation on multiple tasks related to fault analysis in tele-applications, including root-cause analysis, event association prediction, and fault chain tracing, shows that pre-training a language model with tele-domain data is beneficial for downstream tasks. Moreover, the KTeleBERT re-training further improves the performance of task models, highlighting the effectiveness of incorporating diverse tele-knowledge into the model.

cs.AI↗

arXiv2023

DUET: Cross-modal Semantic Grounding for Contrastive Zero-shot Learning

Zero-shot learning (ZSL) aims to predict unseen classes whose samples have never appeared during training. One of the most effective and widely used semantic information for zero-shot image classification are attributes which are annotations for class-level visual characteristics. However, the current methods often fail to discriminate those subtle visual distinctions between images due to not only the shortage of fine-grained annotations, but also the attribute imbalance and co-occurrence. In this paper, we present a transformer-based end-to-end ZSL method named DUET, which integrates latent semantic knowledge from the pre-trained language models (PLMs) via a self-supervised multi-modal learning paradigm. Specifically, we (1) developed a cross-modal semantic grounding network to investigate the model's capability of disentangling semantic attributes from the images; (2) applied an attribute-level contrastive learning strategy to further enhance the model's discrimination on fine-grained visual characteristics against the attribute co-occurrence and imbalance; (3) proposed a multi-task learning policy for considering multi-model objectives. We find that our DUET can achieve state-of-the-art performance on three standard ZSL benchmarks and a knowledge graph equipped ZSL benchmark. Its components are effective and its predictions are interpretable.

cs.CV↗

arXiv2023

Speaker Change Detection for Transformer Transducer ASR

Speaker change detection (SCD) is an important feature that improves the readability of the recognized words from an automatic speech recognition (ASR) system by breaking the word sequence into paragraphs at speaker change points. Existing SCD solutions either require additional ensemble for the time based decisions and recognized word sequences, or implement a tight integration between ASR and SCD, limiting the potential optimum performance for both tasks. To address these issues, we propose a novel framework for the SCD task, where an additional SCD module is built on top of an existing Transformer Transducer ASR (TT-ASR) network. Two variants of the SCD network are explored in this framework that naturally estimate speaker change probability for each word, while allowing the ASR and SCD to have independent optimization scheme for the best performance. Experiments show that our methods can significantly improve the F1 score on LibriCSS and Microsoft call center data sets without ASR degradation, compared with a joint SCD and ASR baseline.

eess.AS↗

arXiv2023

Dipolar Spin Liquid Ending with Quantum Critical Point in a Gd-based Triangular Magnet

By performing experiment and model studies on a triangular-lattice dipolar magnet KBaGd(BO$_3$)$_2$ (KBGB), we find the highly frustrated magnet with a planar anisotropy hosts a strongly fluctuating dipolar spin liquid (DSL), which originates from the intriguing interplay between dipolar and Heisenberg interactions. The DSL constitutes an extended regime in the field-temperature phase diagram, which gets lowered in temperature as field increases and eventually ends with an unconventional quantum critical point (QCP) at $B_c\simeq 0.75$~T. Based on dipolar Heisenberg model calculations, we identify the DSL as a Berezinskii-Kosterlitz-Thouless (BKT) phase with emergent U(1) symmetry. Due to the tremendous entropy accumulation that can be related to the strong BKT and quantum fluctuations, unprecedented magnetic cooling effects are observed in the DSL regime and particularly near the QCP, making KBGB a superior dipolar coolant to commercial Gd-based refrigerants. We establish the phase diagram for triangular-lattice dipolar quantum magnets where emergent symmetry plays an essential role, and provide a basis and opens an avenue for their applications in sub-Kelvin refrigeration.

cond-mat.str-el↗

arXiv2023

Neural Codec Language Models are Zero-Shot Text to Speech Synthesizers

We introduce a language modeling approach for text to speech synthesis (TTS). Specifically, we train a neural codec language model (called Vall-E) using discrete codes derived from an off-the-shelf neural audio codec model, and regard TTS as a conditional language modeling task rather than continuous signal regression as in previous work. During the pre-training stage, we scale up the TTS training data to 60K hours of English speech which is hundreds of times larger than existing systems. Vall-E emerges in-context learning capabilities and can be used to synthesize high-quality personalized speech with only a 3-second enrolled recording of an unseen speaker as an acoustic prompt. Experiment results show that Vall-E significantly outperforms the state-of-the-art zero-shot TTS system in terms of speech naturalness and speaker similarity. In addition, we find Vall-E could preserve the speaker's emotion and acoustic environment of the acoustic prompt in synthesis. See https://aka.ms/valle for demos of our work.

cs.CL↗

arXiv2022

BEATs: Audio Pre-Training with Acoustic Tokenizers

The massive growth of self-supervised learning (SSL) has been witnessed in language, vision, speech, and audio domains over the past few years. While discrete label prediction is widely adopted for other modalities, the state-of-the-art audio SSL models still employ reconstruction loss for pre-training. Compared with reconstruction loss, semantic-rich discrete label prediction encourages the SSL model to abstract the high-level audio semantics and discard the redundant details as in human perception. However, a semantic-rich acoustic tokenizer for general audio pre-training is usually not straightforward to obtain, due to the continuous property of audio and unavailable phoneme sequences like speech. To tackle this challenge, we propose BEATs, an iterative audio pre-training framework to learn Bidirectional Encoder representation from Audio Transformers, where an acoustic tokenizer and an audio SSL model are optimized by iterations. In the first iteration, we use random projection as the acoustic tokenizer to train an audio SSL model in a mask and label prediction manner. Then, we train an acoustic tokenizer for the next iteration by distilling the semantic knowledge from the pre-trained or fine-tuned audio SSL model. The iteration is repeated with the hope of mutual promotion of the acoustic tokenizer and audio SSL model. The experimental results demonstrate our acoustic tokenizers can generate discrete labels with rich audio semantics and our audio SSL models achieve state-of-the-art results across various audio classification benchmarks, even outperforming previous models that use more training data and model parameters significantly. Specifically, we set a new state-of-the-art mAP 50.6% on AudioSet-2M for audio-only models without using any external data, and 98.1% accuracy on ESC-50. The code and pre-trained models are available at https://aka.ms/beats.

eess.AS↗

arXiv2022

Simulating 2+1D Lattice Quantum Electrodynamics at Finite Density with Neural Flow Wavefunctions

We present a neural flow wavefunction, Gauge-Fermion FlowNet, and use it to simulate 2+1D lattice compact quantum electrodynamics with finite density dynamical fermions. The gauge field is represented by a neural network which parameterizes a discretized flow-based transformation of the amplitude while the fermionic sign structure is represented by a neural net backflow. This approach directly represents the $U(1)$ degree of freedom without any truncation, obeys Guass's law by construction, samples autoregressively avoiding any equilibration time, and variationally simulates Gauge-Fermion systems with sign problems accurately. In this model, we investigate confinement and string breaking phenomena in different fermion density and hopping regimes. We study the phase transition from the charge crystal phase to the vacuum phase at zero density, and observe the phase seperation and the net charge penetration blocking effect under magnetic interaction at finite density. In addition, we investigate a magnetic phase transition due to the competition effect between the kinetic energy of fermions and the magnetic energy of the gauge field. With our method, we further note potential differences on the order of the phase transitions between a continuous $U(1)$ system and one with finite truncation. Our state-of-the-art neural network approach opens up new possibilities to study different gauge theories coupled to dynamical matter in higher dimensions.

hep-lat↗

arXiv2022

Zero-shot and Few-shot Learning with Knowledge Graphs: A Comprehensive Survey

Machine learning especially deep neural networks have achieved great success but many of them often rely on a number of labeled samples for supervision. As sufficient labeled training data are not always ready due to e.g., continuously emerging prediction targets and costly sample annotation in real world applications, machine learning with sample shortage is now being widely investigated. Among all these studies, many prefer to utilize auxiliary information including those in the form of Knowledge Graph (KG) to reduce the reliance on labeled samples. In this survey, we have comprehensively reviewed over 90 papers about KG-aware research for two major sample shortage settings -- zero-shot learning (ZSL) where some classes to be predicted have no labeled samples, and few-shot learning (FSL) where some classes to be predicted have only a small number of labeled samples that are available. We first introduce KGs used in ZSL and FSL as well as their construction methods, and then systematically categorize and summarize KG-aware ZSL and FSL methods, dividing them into different paradigms such as the mapping-based, the data augmentation, the propagation-based and the optimization-based. We next present different applications, including not only KG augmented prediction tasks such as image classification, question answering, text classification and knowledge extraction, but also KG completion tasks, and some typical evaluation resources for each task. We eventually discuss some challenges and open problems from different perspectives.

cs.LG↗

arXiv2022

The Star Formation History of the Milky Way's Nuclear Star Cluster

We report the first star formation history study of the Milky Way's nuclear star cluster (NSC) that includes observational constraints from a large sample of stellar metallicity measurements. These metallicity measurements were obtained from recent surveys from Gemini and VLT of 770 late-type stars within the central 1.5 pc. These metallicity measurements, along with photometry and spectroscopically derived temperatures, are forward modeled with a Bayesian inference approach. Including metallicity measurements improves the overall fit quality, as the low-temperature red giants that were previously difficult to constrain are now accounted for, and the best fit favors a two-component model. The dominant component contains 93%$\pm$3% of the mass, is metal-rich ($\overline{[M/H]}\sim$0.45), and has an age of 5$^{+3}_{-2}$ Gyr, which is $\sim$3 Gyr younger than earlier studies with fixed (solar) metallicity; this younger age challenges co-evolutionary models in which the NSC and supermassive black holes formed simultaneously at early times. The minor population component has low metallicity ($\overline{[M/H]}\sim$ -1.1) and contains $\sim$7% of the stellar mass. The age of the minor component is uncertain (0.1 - 5 Gyr old). Using the estimated parameters, we infer the following NSC stellar remnant population (with $\sim$18% uncertainty): 1.5$\times$10$^5$ neutron stars, 2.5$\times$10$^5$ stellar mass black holes (BHs) and 2.2$\times$10$^4$ BH-BH binaries. These predictions result in 2-4 times fewer neutron stars compared to earlier predictions that assume solar metallicity, introducing a possible new path to understand the so-called "missing pulsar problem". Finally, we present updated predictions for the BH-BH merger rates (0.01-3 Gpc$^{-3}$yr$^{-1}$).

astro-ph.GA↗

arXiv2022

LaKo: Knowledge-driven Visual Question Answering via Late Knowledge-to-Text Injection

Visual question answering (VQA) often requires an understanding of visual concepts and language semantics, which relies on external knowledge. Most existing methods exploit pre-trained language models or/and unstructured text, but the knowledge in these resources are often incomplete and noisy. Some other methods prefer to use knowledge graphs (KGs) which often have intensive structured knowledge, but the research is still quite preliminary. In this paper, we propose LaKo, a knowledge-driven VQA method via Late Knowledge-to-text Injection. To effectively incorporate an external KG, we transfer triples into textual format and propose a late injection mechanism for knowledge fusion. Finally we address VQA as a text generation task with an effective encoder-decoder paradigm, which achieves state-of-the-art results on OKVQA dataset.

cs.CV↗

arXiv2022

Speech separation with large-scale self-supervised learning

Self-supervised learning (SSL) methods such as WavLM have shown promising speech separation (SS) results in small-scale simulation-based experiments. In this work, we extend the exploration of the SSL-based SS by massively scaling up both the pre-training data (more than 300K hours) and fine-tuning data (10K hours). We also investigate various techniques to efficiently integrate the pre-trained model with the SS network under a limited computation budget, including a low frame rate SSL model training setup and a fine-tuning scheme using only the part of the pre-trained model. Compared with a supervised baseline and the WavLM-based SS model using feature embeddings obtained with the previously released 94K hours trained WavLM, our proposed model obtains 15.9% and 11.2% of relative word error rate (WER) reductions, respectively, for a simulated far-field speech mixture test set. For conversation transcription on real meeting recordings using continuous speech separation, the proposed model achieves 6.8% and 10.6% of relative WER reductions over the purely supervised baseline on AMI and ICSI evaluation sets, respectively, while reducing the computational cost by 38%.

eess.AS↗

arXiv2022

Exploring WavLM on Speech Enhancement

There is a surge in interest in self-supervised learning approaches for end-to-end speech encoding in recent years as they have achieved great success. Especially, WavLM showed state-of-the-art performance on various speech processing tasks. To better understand the efficacy of self-supervised learning models for speech enhancement, in this work, we design and conduct a series of experiments with three resource conditions by combining WavLM and two high-quality speech enhancement systems. Also, we propose a regression-based WavLM training objective and a noise-mixing data configuration to further boost the downstream enhancement performance. The experiments on the DNS challenge dataset and a simulation dataset show that the WavLM benefits the speech enhancement task in terms of both speech quality and speech recognition accuracy, especially for low fine-tuning resources. For the high fine-tuning resource condition, only the word error rate is substantially improved.

eess.AS↗