arXiv subjects

Chao Zhang

Publications and source records attributed to Chao Zhang.

At least 685 records · Page 38Linked to original sources

arXiv2023

Self-supervised representations in speech-based depression detection

This paper proposes handling training data sparsity in speech-based automatic depression detection (SDD) using foundation models pre-trained with self-supervised learning (SSL). An analysis of SSL representations derived from different layers of pre-trained foundation models is first presented for SDD, which provides insight to suitable indicator for depression detection. Knowledge transfer is then performed from automatic speech recognition (ASR) and emotion recognition to SDD by fine-tuning the foundation models. Results show that the uses of oracle and ASR transcriptions yield similar SDD performance when the hidden representations of the ASR model is incorporated along with the ASR textual information. By integrating representations from multiple foundation models, state-of-the-art SDD results based on real ASR were achieved on the DAIC-WOZ dataset.

cs.CL↗

arXiv2023

Discovery of the high-entropy carbide ceramic topological superconductor candidate (Ti0.2Zr0.2Nb0.2Hf0.2Ta0.2)C

High-entropy ceramics (HECs) are solid solutions of inorganic compounds with one or more Wyckoff sites shared by equal or near-equal atomic ratios of multi-principal elements. Material design and property tailoring possibilities emerge from this new class of materials. Here, we report the discovery of superconductivity around 2.35 K and topological properties in the (Ti0.2Zr0.2Nb0.2Hf0.2Ta0.2)C high-entropy carbide ceramic (HECC), which has not been observed before in any of the investigated HECC. Density functional theory calculations showed that six type-II Dirac points exist in (Ti0.2Zr0.2Nb0.2Hf0.2Ta0.2)C, which mainly contributed from the t2g orbitals of transition metals and the p orbitals of C. Due to the stability of the structure, we also observed robust superconductivity under pressure in this HEC superconductor. This study expands the physical properties of HECs, which may become a new material platform for superconductivity research, especially for studying the coupling between superconductivity and topological physics.

cond-mat.supr-con↗

arXiv2023

Regularity theory for nonlocal equations with general growth in the Heisenberg group

We deal with a wide class of generalized nonlocal $p$-Laplace equations, so-called nonlocal $G$-Laplace equations, in the Heisenberg framework. Under natural hypotheses on the $N$-function $G$, we provide a unified approach to investigate in the spirit of De Giorgi-Nash-Moser theory, some local properties of weak solutions to such kind of problems, involving boundedness, Hölder continuity and Harnack inequality. To this end, an improved nonlocal Caccioppoli-type estimate as the main auxiliary ingredient is exploited several times.

math.AP↗

arXiv2023

Knowledge-Aware Audio-Grounded Generative Slot Filling for Limited Annotated Data

Manually annotating fine-grained slot-value labels for task-oriented dialogue (ToD) systems is an expensive and time-consuming endeavour. This motivates research into slot-filling methods that operate with limited amounts of labelled data. Moreover, the majority of current work on ToD is based solely on text as the input modality, neglecting the additional challenges of imperfect automatic speech recognition (ASR) when working with spoken language. In this work, we propose a Knowledge-Aware Audio-Grounded generative slot-filling framework, termed KA2G, that focuses on few-shot and zero-shot slot filling for ToD with speech input. KA2G achieves robust and data-efficient slot filling for speech-based ToD by 1) framing it as a text generation task, 2) grounding text generation additionally in the audio modality, and 3) conditioning on available external knowledge (e.g. a predefined list of possible slot values). We show that combining both modalities within the KA2G framework improves the robustness against ASR errors. Further, the knowledge-aware slot-value generator in KA2G, implemented via a pointer generator mechanism, particularly benefits few-shot and zero-shot learning. Experiments, conducted on the standard speech-based single-turn SLURP dataset and a multi-turn dataset extracted from a commercial ToD system, display strong and consistent gains over prior work, especially in few-shot and zero-shot setups.

cs.CL↗

arXiv2023

Detecting new fundamental fields with Pulsar Timing Arrays

Strong evidence of the existence of the Stochastic Gravitational-Wave Background (SGWB) has been reported by the NANOGrav, PPTA, EPTA and CPTA collaborations. The Bayesian posteriors of the Gravitational-Wave Background (GWB) amplitude and spectrum are compatible with current astrophysical predictions for the GWB from the population of supermassive black hole binaries (SMBHBs). In this paper, we discuss the corrections arising from the extra scalar or vector radiation to the characteristic dimensionless strain in PTA experiments and explore the possibility to detect charges surrounding massive black holes, which could give rise to SGWB with vector or scalar polarizations. The parametrized frequency-dependent characteristic dimensionless strain is used to take a Bayesian analysis and the Bayes factor is also computed for charged and neutral SMBHBs. The Bayesian posterior of GWB tensor amplitude is $\log_{10} A_T=-14.85^{+0.26}_{-0.38}$ and spectral exponent $α=-0.60^{+0.32}_{-0.36}$. The Bayesian posterior for vector or scalar amplitude $A_{V, S}$ is nearly flat and there is nearly no constraint from the current observation data. The Bayesian factor is $0.71$ far less than 100, so the current observation can not support the existence of the charged SMBHB.

gr-qc↗

arXiv2023

CDMA: A Practical Cross-Device Federated Learning Algorithm for General Minimax Problems

Minimax problems arise in a wide range of important applications including robust adversarial learning and Generative Adversarial Network (GAN) training. Recently, algorithms for minimax problems in the Federated Learning (FL) paradigm have received considerable interest. Existing federated algorithms for general minimax problems require the full aggregation (i.e., aggregation of local model information from all clients) in each training round. Thus, they are inapplicable to an important setting of FL known as the cross-device setting, which involves numerous unreliable mobile/IoT devices. In this paper, we develop the first practical algorithm named CDMA for general minimax problems in the cross-device FL setting. CDMA is based on a Start-Immediately-With-Enough-Responses mechanism, in which the server first signals a subset of clients to perform local computation and then starts to aggregate the local results reported by clients once it receives responses from enough clients in each round. With this mechanism, CDMA is resilient to the low client availability. In addition, CDMA is incorporated with a lightweight global correction in the local update steps of clients, which mitigates the impact of slow network connections. We establish theoretical guarantees of CDMA under different choices of hyperparameters and conduct experiments on AUC maximization, robust adversarial network training, and GAN training tasks. Theoretical and experimental results demonstrate the efficiency of CDMA.

cs.LG↗

arXiv2023

Towards Optimal Randomized Strategies in Adversarial Example Game

The vulnerability of deep neural network models to adversarial example attacks is a practical challenge in many artificial intelligence applications. A recent line of work shows that the use of randomization in adversarial training is the key to find optimal strategies against adversarial example attacks. However, in a fully randomized setting where both the defender and the attacker can use randomized strategies, there are no efficient algorithm for finding such an optimal strategy. To fill the gap, we propose the first algorithm of its kind, called FRAT, which models the problem with a new infinite-dimensional continuous-time flow on probability distribution spaces. FRAT maintains a lightweight mixture of models for the defender, with flexibility to efficiently update mixing weights and model parameters at each iteration. Furthermore, FRAT utilizes lightweight sampling subroutines to construct a random strategy for the attacker. We prove that the continuous-time limit of FRAT converges to a mixed Nash equilibria in a zero-sum game formed by a defender and an attacker. Experimental results also demonstrate the efficiency of FRAT on CIFAR-10 and CIFAR-100 datasets.

cs.LG↗

arXiv2023

Detecting vector charge with extreme mass ratio inspirals onto Kerr black holes

Extreme mass ratio inspirals (EMRIs) are excellent sources for space-based observatories to explore the properties of black holes and test no-hair theorems. We consider EMRIs with a charged compact object inspiralling onto a Kerr black hole in quasi-circular orbits. Using the Teukolsky and generalized Sasaki-Nakamura formalisms for the gravitational and vector perturbations about a Kerr black hole, we numerically calculate the energy fluxes for both gravitational and vector perturbations induced by a charged particle moving in equatorial circular orbits. With one-year observations of EMRIs, we apply the Fisher information matrix method to estimate the charge uncertainty detected by space-based gravitational wave detectors such as the Laser Interferometer Space Antenna, TianQin, and Taiji, and we find that it is possible to detect vector charge as small as $q\sim 0.0049$. The results show that EMRIs composed of a Kerr black hole with a higher spin $a$ and lighter mass $M$, and a secondary charged object with more vector charge give smaller relative error on the charge, thus constrain the charge better. The positive spin of the Kerr black hole can decrease the charge uncertainty by about one or two orders of magnitude.

gr-qc↗

arXiv2023

Gravitational waves from eccentric extreme mass-ratio inspirals as probes of scalar fields

We study eccentric orbits of the Schwarzschild spacetime for extreme mass ratio system (EMRI) in modified gravity theories with additional scalar fields. Due to the additional energy and angular momentum carried away by the scalar field, the orbit of the EMRI in modified gravity decays faster than that in general relativity. The time that it takes the eccentricity $e$ to reach the minimum is smaller and the values of the semi-latus rectum $p$ and $e$ at the turning point when $e$ reaches the minimum are bigger for larger scalar charge $d$. In addition to the calculation of energy fluxes with numerical method, we also use the Post-Newtonian expansion of the rate of energy carried away by the scalar field in eccentric orbits to understand the behaviors of the energy emission. By adding the scalar flux to the open code FastEMRIWaveforms of the Black Hole Perturbation Toolkit, we numerically generate fast gravitational waveforms for eccentric EMRIs with scalar fields and use the faithfulness between waveforms with and without the scalar charge to discuss the detection of scalar charge $d$. The detection error of the scalar charge is also estimated with method of Fisher information matrix.

gr-qc↗

arXiv2023

A Solution to CVPR'2023 AQTC Challenge: Video Alignment for Multi-Step Inference

Affordance-centric Question-driven Task Completion (AQTC) for Egocentric Assistant introduces a groundbreaking scenario. In this scenario, through learning instructional videos, AI assistants provide users with step-by-step guidance on operating devices. In this paper, we present a solution for enhancing video alignment to improve multi-step inference. Specifically, we first utilize VideoCLIP to generate video-script alignment features. Afterwards, we ground the question-relevant content in instructional videos. Then, we reweight the multimodal context to emphasize prominent features. Finally, we adopt GRU to conduct multi-step inference. Through comprehensive experiments, we demonstrate the effectiveness and superiority of our method, which secured the 2nd place in CVPR'2023 AQTC challenge. Our code is available at https://github.com/zcfinal/LOVEU-CVPR23-AQTC.

cs.CV↗

arXiv2023

Regularity of weak solutions for mixed local and nonlocal double phase parabolic equations

We study the mixed local and nonlocal double phase parabolic equation \begin{align*} \partial_t u(x,t)-\mathrm{div}(a(x,t)|\nabla u|^{q-2}\nabla u) +\mathcal{L}u(x,t)=0 \end{align*} in $Q_T=Ω\times(0,T)$, where $\mathcal{L}$ is the nonlocal $p$-Laplace type operator. The local boundedness of weak solutions is proved by means of the De Giorgi-Nash-Moser iteration with the nonnegative coefficient function $a(x,t)$ being bounded. In addition, when $a(x,t)$ is Hölder continuous, we discuss the pointwise behavior and lower semicontinuity of weak supersolutions based on energy estimates and De Giorgi type lemma. Analogously, the corresponding results are also valid for weak subsolutions.

math.AP↗

arXiv2023

G-STO: Sequential Main Shopping Intention Detection via Graph-Regularized Stochastic Transformer

Sequential recommendation requires understanding the dynamic patterns of users' behaviors, contexts, and preferences from their historical interactions. Most existing works focus on modeling user-item interactions only from the item level, ignoring that they are driven by latent shopping intentions (e.g., ballpoint pens, miniatures, etc). The detection of the underlying shopping intentions of users based on their historical interactions is a crucial aspect for e-commerce platforms, such as Amazon, to enhance the convenience and efficiency of their customers' shopping experiences. Despite its significance, the area of main shopping intention detection remains under-investigated in the academic literature. To fill this gap, we propose a graph-regularized stochastic Transformer method, G-STO. By considering intentions as sets of products and user preferences as compositions of intentions, we model both of them as stochastic Gaussian embeddings in the latent representation space. Instead of training the stochastic representations from scratch, we develop a global intention relational graph as prior knowledge for regularization, allowing relevant shopping intentions to be distributionally close. Finally, we feed the newly regularized stochastic embeddings into Transformer-based models to encode sequential information from the intention transitions. We evaluate our main shopping intention identification model on three different real-world datasets, where G-STO achieves significantly superior performances to the baselines by 18.08% in Hit@1, 7.01% in Hit@10, and 6.11% in NDCG@10 on average.

cs.IR↗

arXiv2023

ToolQA: A Dataset for LLM Question Answering with External Tools

Large Language Models (LLMs) have demonstrated impressive performance in various NLP tasks, but they still suffer from challenges such as hallucination and weak numerical reasoning. To overcome these challenges, external tools can be used to enhance LLMs' question-answering abilities. However, current evaluation methods do not distinguish between questions that can be answered using LLMs' internal knowledge and those that require external information through tool use. To address this issue, we introduce a new dataset called ToolQA, which is designed to faithfully evaluate LLMs' ability to use external tools for question answering. Our development of ToolQA involved a scalable, automated process for dataset curation, along with 13 specialized tools designed for interaction with external knowledge in order to answer questions. Importantly, we strive to minimize the overlap between our benchmark data and LLMs' pre-training data, enabling a more precise evaluation of LLMs' tool-use reasoning abilities. We conducted an in-depth diagnosis of existing tool-use LLMs to highlight their strengths, weaknesses, and potential improvements. Our findings set a new benchmark for evaluating LLMs and suggest new directions for future advancements. Our data and code are freely available to the broader scientific community on GitHub.

cs.CL↗

arXiv2023

Pushing the Limits of Unsupervised Unit Discovery for SSL Speech Representation

The excellent generalization ability of self-supervised learning (SSL) for speech foundation models has garnered significant attention. HuBERT is a successful example that utilizes offline clustering to convert speech features into discrete units for a masked language modeling pretext task. However, simply clustering features as targets by k-means does not fully inspire the model's performance. In this work, we present an unsupervised method to improve SSL targets. Two models are proposed, MonoBERT and PolyBERT, which leverage context-independent and context-dependent phoneme-based units for pre-training. Our models outperform other SSL models significantly on the LibriSpeech benchmark without the need for iterative re-clustering and re-training. Furthermore, our models equipped with context-dependent units even outperform target-improvement models that use labeled data during pre-training. How we progressively improve the unit discovery process is demonstrated through experiments.

cs.CL↗

arXiv2023

Cross-Impact of Order Flow Imbalance in Equity Markets

We investigate the impact of order flow imbalance (OFI) on price movements in equity markets in a multi-asset setting. First, we propose a systematic approach for combining OFIs at the top levels of the limit order book into an integrated OFI variable which better explains price impact, compared to the best-level OFI. We show that once the information from multiple levels is integrated into OFI, multi-asset models with cross-impact do not provide additional explanatory power for contemporaneous impact compared to a sparse model without cross-impact terms. On the other hand, we show that lagged cross-asset OFIs do improve the forecasting of future returns. We also establish that this lagged cross-impact mainly manifests at short-term horizons and decays rapidly in time.

q-fin.TR↗

arXiv2023

DyGen: Learning from Noisy Labels via Dynamics-Enhanced Generative Modeling

Learning from noisy labels is a challenge that arises in many real-world applications where training data can contain incorrect or corrupted labels. When fine-tuning language models with noisy labels, models can easily overfit the label noise, leading to decreased performance. Most existing methods for learning from noisy labels use static input features for denoising, but these methods are limited by the information they can provide on true label distributions and can result in biased or incorrect predictions. In this work, we propose the Dynamics-Enhanced Generative Model (DyGen), which uses dynamic patterns in the embedding space during the fine-tuning process of language models to improve noisy label predictions. DyGen uses the variational auto-encoding framework to infer the posterior distributions of true labels from noisy labels and training dynamics. Additionally, a co-regularization mechanism is used to minimize the impact of potentially noisy labels and priors. DyGen demonstrates an average accuracy improvement of 3.10% on two synthetic noise datasets and 1.48% on three real-world noise datasets compared to the previous state-of-the-art. Extensive experiments and analyses show the effectiveness of each component in DyGen. Our code is available for reproducibility on GitHub.

cs.CL↗

arXiv2023

Model-Aware Contrastive Learning: Towards Escaping the Dilemmas

Contrastive learning (CL) continuously achieves significant breakthroughs across multiple domains. However, the most common InfoNCE-based methods suffer from some dilemmas, such as \textit{uniformity-tolerance dilemma} (UTD) and \textit{gradient reduction}, both of which are related to a $\mathcal{P}_{ij}$ term. It has been identified that UTD can lead to unexpected performance degradation. We argue that the fixity of temperature is to blame for UTD. To tackle this challenge, we enrich the CL loss family by presenting a Model-Aware Contrastive Learning (MACL) strategy, whose temperature is adaptive to the magnitude of alignment that reflects the basic confidence of the instance discrimination task, then enables CL loss to adjust the penalty strength for hard negatives adaptively. Regarding another dilemma, the gradient reduction issue, we derive the limits of an involved gradient scaling factor, which allows us to explain from a unified perspective why some recent approaches are effective with fewer negative samples, and summarily present a gradient reweighting to escape this dilemma. Extensive remarkable empirical results in vision, sentence, and graph modality validate our approach's general improvement for representation learning and downstream tasks.

cs.LG↗

arXiv2023

Estimating the Uncertainty in Emotion Attributes using Deep Evidential Regression

In automatic emotion recognition (AER), labels assigned by different human annotators to the same utterance are often inconsistent due to the inherent complexity of emotion and the subjectivity of perception. Though deterministic labels generated by averaging or voting are often used as the ground truth, it ignores the intrinsic uncertainty revealed by the inconsistent labels. This paper proposes a Bayesian approach, deep evidential emotion regression (DEER), to estimate the uncertainty in emotion attributes. Treating the emotion attribute labels of an utterance as samples drawn from an unknown Gaussian distribution, DEER places an utterance-specific normal-inverse gamma prior over the Gaussian likelihood and predicts its hyper-parameters using a deep neural network model. It enables a joint estimation of emotion attributes along with the aleatoric and epistemic uncertainties. AER experiments on the widely used MSP-Podcast and IEMOCAP datasets showed DEER produced state-of-the-art results for both the mean values and the distribution of emotion attributes.

cs.SD↗