arXiv subjects

Jian Kang

Publications and source records attributed to Jian Kang.

At least 55 records · Page 3Linked to original sources

arXiv2025

DIFFA: Large Language Diffusion Models Can Listen and Understand

Recent advances in large language models (LLMs) have shown remarkable capabilities across textual and multimodal domains. In parallel, diffusion-based language models have emerged as a promising alternative to the autoregressive paradigm, offering improved controllability, bidirectional context modeling, and robust generation. However, their application to the audio modality remains underexplored. In this work, we introduce \textbf{DIFFA}, the first diffusion-based large audio-language model designed to perform spoken language understanding. DIFFA integrates a frozen diffusion language model with a lightweight dual-adapter architecture that bridges speech understanding and natural language reasoning. We employ a two-stage training pipeline: first, aligning semantic representations via an ASR objective; then, learning instruction-following abilities through synthetic audio-caption pairs automatically generated by prompting LLMs. Despite being trained on only 960 hours of ASR and 127 hours of synthetic instruction data, DIFFA demonstrates competitive performance on major benchmarks, including MMSU, MMAU, and VoiceBench, outperforming several autoregressive open-source baselines. Our results reveal the potential of diffusion-based language models for efficient and scalable audio understanding, opening a new direction for speech-driven AI. Our code will be available at https://github.com/NKU-HLT/DIFFA.git.

cs.SD↗

arXiv2025

Bridging the Gap between Continuous and Informative Discrete Representations by Random Product Quantization

Self-supervised learning (SSL) has become a core technique in speech processing, but the high dimensionality of its representations makes discretization essential for improving efficiency. However, existing discretization methods still suffer from significant information loss, resulting in a notable performance gap compared to continuous representations. To overcome these limitations, we propose two quantization-based discretization methods: Product Quantization (PQ) and Random Product Quantization (RPQ). PQ partitions the original feature space into multiple subspaces and independently quantizes each sub-vector, producing a fused set of discrete units that retain diverse information from different subspaces, thereby mitigating the loss associated with single-cluster quantization. RPQ further enhances representation diversity by randomly sampling a fixed proportion of feature dimensions multiple times to construct sub-vectors, thereby better capturing the variability in the data distribution. Theoretical analysis shows that RPQ reduces the correlation coefficient rho (where 0 <= rho <= 1) between sub-quantizers. Its quantization error is lower-bounded by the product of rho and epsilon-kms, where epsilon-kms denotes the quantization error of a single K-means quantizer. Experimental results on a combined dataset built from LibriSpeech and ML-SUPERB show that PQ and RPQ outperform standard K-means discretization, achieving relative improvements of 21.8 percent and 20.0 percent in WER on LibriSpeech, and 24.1 percent and 19.6 percent in CER on ML-SUPERB, respectively. Moreover, their performance is competitive with, and in some cases even surpasses, that of continuous SSL representations.

eess.AS↗

arXiv2025

CLIMB: Class-imbalanced Learning Benchmark on Tabular Data

Class-imbalanced learning (CIL) on tabular data is important in many real-world applications where the minority class holds the critical but rare outcomes. In this paper, we present CLIMB, a comprehensive benchmark for class-imbalanced learning on tabular data. CLIMB includes 73 real-world datasets across diverse domains and imbalance levels, along with unified implementations of 29 representative CIL algorithms. Built on a high-quality open-source Python package with unified API designs, detailed documentation, and rigorous code quality controls, CLIMB supports easy implementation and comparison between different CIL algorithms. Through extensive experiments, we provide practical insights on method accuracy and efficiency, highlighting the limitations of naive rebalancing, the effectiveness of ensembles, and the importance of data quality. Our code, documentation, and examples are available at https://github.com/ZhiningLiu1998/imbalanced-ensemble.

cs.LG↗

arXiv2025

Bayesian Signal Matching for Transfer Learning in ERP-Based Brain Computer Interface

An Event-Related Potential (ERP)-based Brain-Computer Interface (BCI) Speller System assists people with disabilities to communicate by decoding electroencephalogram (EEG) signals. A P300-ERP embedded in EEG signals arises in response to a rare, but relevant event (target) among a series of irrelevant events (non-target). Different machine learning methods have constructed binary classifiers to detect target events, known as calibration. The existing calibration strategy uses data from participants themselves with lengthy training time. Participants feel bored and distracted, which causes biased P300 estimation and decreased prediction accuracy. To resolve this issue, we propose a Bayesian signal matching (BSM) framework to calibrate EEG signals from a new participant using data from source participants. BSM specifies the joint distribution of stimulus-specific EEG signals among source participants via a Bayesian hierarchical mixture model. We apply the inference strategy. If source and new participants are similar, they share the same set of model parameters; otherwise, they keep their own sets of model parameters; we predict on the testing data using parameters of the baseline cluster directly. Our hierarchical framework can be generalized to other base classifiers with parametric forms. We demonstrate the advantages of BSM using simulations and focus on the real data analysis among participants with neuro-degenerative diseases.

stat.AP↗

arXiv2025

Analytical solution for the relaxed atomic configuration of twisted bilayer graphene including heterostrain

Continuum atomic relaxation models for twisted bilayer graphene involve minimization of the sum of intralayer elastic energy and interlayer adhesion energy. The elastic energy favors a rigid twist i.e. no distortion in the twisted honeycomb lattices, while the adhesion energy favors Bernal stacking and breaking the relaxation into triangular AB and BA stacked domains. We compare the results of two relaxation models with the published Bragg interferometry data, finding good agreement with one of the models. We then provide a method for finding a highly accurate approximation to the solution of this model which holds above the twist angle of $\approx0.7^\circ$ and thus covers the first magic angle. We find closed form expressions in the absence, as well as in the presence, of external heterostrain. These expressions are not written as a Taylor series in the ratio of adhesion and elastic energy, because, as we show, the radius of convergence of such a series is too small to access the first magic angle.

cond-mat.str-el↗

arXiv2025

Analyzing Uncertainty of LLM-as-a-Judge: Interval Evaluations with Conformal Prediction

LLM-as-a-judge has become a promising paradigm for using large language models (LLMs) to evaluate natural language generation (NLG), but the uncertainty of its evaluation remains underexplored. This lack of reliability may limit its deployment in many applications. This work presents the first framework to analyze the uncertainty by offering a prediction interval of LLM-based scoring via conformal prediction. Conformal prediction constructs continuous prediction intervals from a single evaluation run, and we design an ordinal boundary adjustment for discrete rating tasks. We also suggest a midpoint-based score within the interval as a low-bias alternative to raw model score and weighted average. We perform extensive experiments and analysis, which show that conformal prediction can provide valid prediction interval with coverage guarantees. We also explore the usefulness of interval midpoint and judge reprompting for better judgment.

cs.CL↗

arXiv2025

WenetSpeech-Chuan: A Large-Scale Sichuanese Corpus with Rich Annotation for Dialectal Speech Processing

The scarcity of large-scale, open-source data for dialects severely hinders progress in speech technology, a challenge particularly acute for the widely spoken Sichuanese dialects of Chinese. To address this critical gap, we introduce WenetSpeech-Chuan, a 10,000-hour, richly annotated corpus constructed using our novel Chuan-Pipeline, a complete data processing framework for dialectal speech. To facilitate rigorous evaluation and demonstrate the corpus's effectiveness, we also release high-quality ASR and TTS benchmarks, WenetSpeech-Chuan-Eval, with manually verified transcriptions. Experiments show that models trained on WenetSpeech-Chuan achieve state-of-the-art performance among open-source systems and demonstrate results comparable to commercial services. As the largest open-source corpus for Sichuanese dialects, WenetSpeech-Chuan not only lowers the barrier to research in dialectal speech processing but also plays a crucial role in promoting AI equity and mitigating bias in speech technologies. The corpus, benchmarks, models, and receipts are publicly available on our project page.

cs.CL↗

arXiv2025

Bayesian Image-on-Image Regression via Deep Kernel Learning based Gaussian Processes

In neuroimaging studies, it becomes increasingly important to study associations between different imaging modalities using image-on-image regression (IIR), which faces challenges in interpretation, statistical inference, and prediction. Our motivating problem is how to predict task-evoked fMRI activity using resting-state fMRI data in the Human Connectome Project (HCP). The main difficulty lies in effectively combining different types of imaging predictors with varying resolutions and spatial domains in IIR. To address these issues, we develop Bayesian Image-on-image Regression via Deep Kernel Learning Gaussian Processes (BIRD-GP) and develop efficient posterior computation methods through Stein variational gradient descent. We demonstrate the advantages of BIRD-GP over state-of-the-art IIR methods using simulations. For HCP data analysis using BIRD-GP, we combine the voxel-wise fALFF maps and region-wise connectivity matrices to predict fMRI contrast maps for language and social recognition tasks. We show that fALFF is less predictive than the connectivity matrix for both tasks, but combining both yields improved results. Angular Gyrus Right emerges as the most predictable region for the language task (75.9% predictable voxels), while Superior Parietal Gyrus Right tops for the social recognition task (48.9% predictable voxels). Additionally, we identify features from the resting-state fMRI data that are important for task fMRI prediction.

stat.AP↗

arXiv2025

Scalable Bayesian inference for heat kernel Gaussian processes on manifolds

We establish a scalable manifold learning method and theory, motivated by the problem of estimating fMRI activation manifolds in the Human Connectome Project (HCP). Our primary contribution is the development of an efficient estimation technique for heat kernel Gaussian processes in the exponential family model. This approach handles large sample sizes $n$, preserves the intrinsic geometry of data, and significantly reduces computational complexity from $\mathcal{O}(n^3)$ to $\mathcal{O}(n)$ via a novel reduced-rank approximation of the graph Laplacian's transition matrix and a Truncated Singular Value Decomposition for the eigenpair computation. The numerical experiments demonstrate the scalability and improved accuracy of our method for manifold learning tasks involving complex large-scale data.

stat.ME↗

arXiv2025

WenetSpeech-Yue: A Large-scale Cantonese Speech Corpus with Multi-dimensional Annotation

The development of speech understanding and generation has been significantly accelerated by the availability of large-scale, high-quality speech datasets. Among these, ASR and TTS are regarded as the most established and fundamental tasks. However, for Cantonese (Yue Chinese), spoken by approximately 84.9 million native speakers worldwide, limited annotated resources have hindered progress and resulted in suboptimal ASR and TTS performance. To address this challenge, we propose WenetSpeech-Pipe, an integrated pipeline for building large-scale speech corpus with multi-dimensional annotation tailored for speech understanding and generation. It comprises six modules: Audio Collection, Speaker Attributes Annotation, Speech Quality Annotation, Automatic Speech Recognition, Text Postprocessing and Recognizer Output Voting, enabling rich and high-quality annotations. Based on this pipeline, we release WenetSpeech-Yue, the first large-scale Cantonese speech corpus with multi-dimensional annotation for ASR and TTS, covering 21,800 hours across 10 domains with annotations including ASR transcription, text confidence, speaker identity, age, gender, speech quality scores, among other annotations. We also release WSYue-eval, a comprehensive Cantonese benchmark with two components: WSYue-ASR-eval, a manually annotated set for evaluating ASR on short and long utterances, code-switching, and diverse acoustic conditions, and WSYue-TTS-eval, with base and coverage subsets for standard and generalization testing. Experimental results show that models trained on WenetSpeech-Yue achieve competitive results against state-of-the-art (SOTA) Cantonese ASR and TTS systems, including commercial and LLM-based models, highlighting the value of our dataset and pipeline.

cs.SD↗

arXiv2025

Multilingual Speech Recognition Using Discrete Tokens with a Two-step Training Strategy

Pre-trained models, especially self-supervised learning (SSL) models, have demonstrated impressive results in automatic speech recognition (ASR) task. While most applications of SSL models focus on leveraging continuous representations as features for training downstream tasks, the utilization of discrete units has gained increasing attention in recent years owing to its lower storage requirements and broader range of applications. In multilingual ASR tasks, representations at different layers of the model contribute differently to various languages, complicating the unification of discrete unit modeling. In this paper, we propose a two-stage training strategy to improve the discrete token performance of pre-trained models and narrow the gap with continuous representation performance. We validate our method on the XLS-R model following the settings of Interspeech2024 Speech Processing Using Discrete Speech Unit Challenge. Our method demonstrates a significant improvement on the ML-SUPERB dataset, achieving a 44% relative reduction on CER for the XLS-R model. This surpasses the previous baseline set by the WavLM model, which achieves a 26% relative reduction on CER. Furthermore, our method achieves the first place among all the single-system results on the leaderboard.

eess.AS↗

arXiv2025

Soundscape Captioning using Sound Affective Quality Network and Large Language Model

We live in a rich and varied acoustic world, which is experienced by individuals or communities as a soundscape. Computational auditory scene analysis, disentangling acoustic scenes by detecting and classifying events, focuses on objective attributes of sounds, such as their category and temporal characteristics, ignoring their effects on people, such as the emotions they evoke within a context. To fill this gap, we propose the affective soundscape captioning (ASSC) task, which enables automated soundscape analysis, thus avoiding labour-intensive subjective ratings and surveys in conventional methods. With soundscape captioning, context-aware descriptions are generated for soundscape by capturing the acoustic scenes (ASs), audio events (AEs) information, and the corresponding human affective qualities (AQs). To this end, we propose an automatic soundscape captioner (SoundSCaper) system composed of an acoustic model, i.e. SoundAQnet, and a large language model (LLM). SoundAQnet simultaneously models multi-scale information about ASs, AEs, and perceived AQs, while the LLM describes the soundscape with captions by parsing the information captured with SoundAQnet. SoundSCaper is assessed by two juries of 32 people. In expert evaluation, the average score of SoundSCaper-generated captions is slightly lower than that of two soundscape experts on the evaluation set D1 and the external mixed dataset D2, but not statistically significant. In layperson evaluation, SoundSCaper outperforms soundscape experts in several metrics. In addition to human evaluation, compared to other automated audio captioning systems with and without LLM, SoundSCaper performs better on the ASSC task in several NLP-based metrics. Overall, SoundSCaper performs well in human subjective evaluation and various objective captioning metrics, and the generated captions are comparable to those annotated by soundscape experts.

eess.AS↗

arXiv2025

EVINET: Towards Open-World Graph Learning via Evidential Reasoning Network

Graph learning has been crucial to many real-world tasks, but they are often studied with a closed-world assumption, with all possible labels of data known a priori. To enable effective graph learning in an open and noisy environment, it is critical to inform the model users when the model makes a wrong prediction to in-distribution data of a known class, i.e., misclassification detection or when the model encounters out-of-distribution from novel classes, i.e., out-of-distribution detection. This paper introduces Evidential Reasoning Network (EVINET), a framework that addresses these two challenges by integrating Beta embedding within a subjective logic framework. EVINET includes two key modules: Dissonance Reasoning for misclassification detection and Vacuity Reasoning for out-of-distribution detection. Extensive experiments demonstrate that EVINET outperforms state-of-the-art methods across multiple metrics in the tasks of in-distribution classification, misclassification detection, and out-of-distribution detection. EVINET demonstrates the necessity of uncertainty estimation and logical reasoning for misclassification detection and out-of-distribution detection and paves the way for open-world graph learning. Our code and data are available at https://github.com/SSSKJ/EviNET.

cs.LG↗

arXiv2025

GOAT-SLM: A Spoken Language Model with Paralinguistic and Speaker Characteristic Awareness

Recent advances in end-to-end spoken language models (SLMs) have significantly improved the ability of AI systems to engage in natural spoken interactions. However, most existing models treat speech merely as a vehicle for linguistic content, often overlooking the rich paralinguistic and speaker characteristic cues embedded in human speech, such as dialect, age, emotion, and non-speech vocalizations. In this work, we introduce GOAT-SLM, a novel spoken language model with paralinguistic and speaker characteristic awareness, designed to extend spoken language modeling beyond text semantics. GOAT-SLM adopts a dual-modality head architecture that decouples linguistic modeling from acoustic realization, enabling robust language understanding while supporting expressive and adaptive speech generation. To enhance model efficiency and versatility, we propose a modular, staged training strategy that progressively aligns linguistic, paralinguistic, and speaker characteristic information using large-scale speech-text corpora. Experimental results on TELEVAL, a multi-dimensional evaluation benchmark, demonstrate that GOAT-SLM achieves well-balanced performance across both semantic and non-semantic tasks, and outperforms existing open-source models in handling emotion, dialectal variation, and age-sensitive interactions. This work highlights the importance of modeling beyond linguistic content and advances the development of more natural, adaptive, and socially aware spoken language systems.

cs.CL↗

arXiv2025

BoSS: Beyond-Semantic Speech

Human communication involves more than explicit semantics, with implicit signals and contextual cues playing a critical role in shaping meaning. However, modern speech technologies, such as Automatic Speech Recognition (ASR) and Text-to-Speech (TTS) often fail to capture these beyond-semantic dimensions. To better characterize and benchmark the progression of speech intelligence, we introduce Spoken Interaction System Capability Levels (L1-L5), a hierarchical framework illustrated the evolution of spoken dialogue systems from basic command recognition to human-like social interaction. To support these advanced capabilities, we propose Beyond-Semantic Speech (BoSS), which refers to the set of information in speech communication that encompasses but transcends explicit semantics. It conveys emotions, contexts, and modifies or extends meanings through multidimensional features such as affective cues, contextual dynamics, and implicit semantics, thereby enhancing the understanding of communicative intentions and scenarios. We present a formalized framework for BoSS, leveraging cognitive relevance theories and machine learning models to analyze temporal and contextual speech dynamics. We evaluate BoSS-related attributes across five different dimensions, reveals that current spoken language models (SLMs) are hard to fully interpret beyond-semantic signals. These findings highlight the need for advancing BoSS research to enable richer, more context-aware human-machine communication.

cs.SD↗

arXiv2025

Extended Fractional Chern Insulators Near Half Flux in Twisted Bilayer Graphene Above the Magic Angle

Fractional Chern insulators (FCIs) -- the lattice analog of fractional quantum Hall states -- form as fractionalized quasiparticles emerge in a partially-filled Chern band. This fractionalization is driven by the interplay of electronic interaction and quantum geometry of the underlying wavefunctions. Bilayer graphene with an interlayer twist near the magic angle of 1.1\textdegree\ hosts diverse correlated electronic states at zero magnetic field. When the twist angle exceeds 1.3\textdegree, the electronic bandwidth is sufficient to suppress the zero-field correlated states. Yet applying a magnetic field can restore the importance of electron-electron interactions. Here, we report strongly-correlated phases when a 1.37\textdegree\ twisted bilayer graphene sample is tuned to near half a magnetic flux quantum per moiré cell, deep into the Hofstadter regime. Most notably, well-quantized odd-denominator FCI states appear in multiple Hofstadter subbands over unusually large ranges of density. We also observe a bending and resetting of the Landau minifan reminiscent of behavior commonly seen in magic-angle samples near integer filling at low magnetic field.

cond-mat.mes-hall↗

arXiv2025

Non-exchangeable Conformal Prediction for Temporal Graph Neural Networks

Conformal prediction for graph neural networks (GNNs) offers a promising framework for quantifying uncertainty, enhancing GNN reliability in high-stakes applications. However, existing methods predominantly focus on static graphs, neglecting the evolving nature of real-world graphs. Temporal dependencies in graph structure, node attributes, and ground truth labels violate the fundamental exchangeability assumption of standard conformal prediction methods, limiting their applicability. To address these challenges, in this paper, we introduce NCPNET, a novel end-to-end conformal prediction framework tailored for temporal graphs. Our approach extends conformal prediction to dynamic settings, mitigating statistical coverage violations induced by temporal dependencies. To achieve this, we propose a diffusion-based non-conformity score that captures both topological and temporal uncertainties within evolving networks. Additionally, we develop an efficiency-aware optimization algorithm that improves the conformal prediction process, enhancing computational efficiency and reducing coverage violations. Extensive experiments on diverse real-world temporal graphs, including WIKI, REDDIT, DBLP, and IBM Anti-Money Laundering dataset, demonstrate NCPNET's capability to ensure guaranteed coverage in temporal graphs, achieving up to a 31% reduction in prediction set size on the WIKI dataset, significantly improving efficiency compared to state-of-the-art methods. Our data and code are available at https://github.com/ODYSSEYWT/NCPNET.

cs.LG↗

arXiv2025

Light-induced Pairing Instability of Ultrafast Electron Beams with Space Charge Interactions

Ultrafast electron beams are essential for many applications, yet space-charge interactions in high-intensity beams lead to energy dissipation, coherence loss, and pulse broadening. Existing techniques mitigate these effects by using low-flux beams, preserving beam coherence into the quantum regime. Here, we propose a novel approach by treating the electrons as a strongly correlated Fermi gas rather than merely as an ensemble of charged point-like particles. We introduce a photon-induced pairing mechanism that generates a net attractive force between two electrons, thereby forming "flying bound states" analogous to Cooper pairs of conduction electrons in superconductors. Employing the setting of photon-induced near-field electron microscopy (PINEM), we demonstrate that the effective interaction via single-photon exchange among PINEM electrons can suppress the inherent repulsive Coulomb interaction, enabling a pairing instability mediated by structured electromagnetic fields at near-resonant velocity matching regimes. Finally, we analyze the dynamics of the free-electron pairs in a bunched beam, underscoring the potential to facilitate a phase-coherent condensate of electrons, which can further enhance beam coherence and multi-particle correlation for high-intensity electrons.

physics.optics↗