arXiv subjects

Yifan Hu

Publications and source records attributed to Yifan Hu.

At least 73 records · Page 4Linked to original sources

arXiv2025

Entropy-Driven Pre-Tokenization for Byte-Pair Encoding

Byte-Pair Encoding (BPE) has become a widely adopted subword tokenization method in modern language models due to its simplicity and strong empirical performance across downstream tasks. However, applying BPE to unsegmented languages such as Chinese presents significant challenges, as its frequency-driven merge operation is agnostic to linguistic boundaries. To address this, we propose two entropy-informed pre-tokenization strategies that guide BPE segmentation using unsupervised information-theoretic cues. The first approach uses pointwise mutual information and left/right entropy to identify coherent character spans, while the second leverages predictive entropy derived from a pretrained GPT-2 model to detect boundary uncertainty. We evaluate both methods on a subset of the PKU dataset and demonstrate substantial improvements in segmentation precision, recall, and F1 score compared to standard BPE. Our results suggest that entropy-guided pre-tokenization not only enhances alignment with gold-standard linguistic units but also offers a promising direction for improving tokenization quality in low-resource and multilingual settings.

cs.CL↗

arXiv2025

GANORM: Lifespan Normative Modeling of EEG Network Topology based on Multinational Cross-Spectra

Charting the lifespan evolutionary trajectory of brain function serves as the normative standard for preventing mental disorders during brain development and aging. Although numerous MRI studies have mapped the structural connectome for young cohorts, the EEG-based functional connectome is unknown to characterize human lifespan, limiting its practical applications for the early detection of brain dysfunctions at the community level. This work aimed to undertake normative modeling from the perspective of EEG network topology. Frequency-dependent scalp EEG functional networks were constructed based on EEG cross-spectra aged 5-97 years from 9 countries and network characteristics were quantified. First, GAMLSS were applied to describe the normative curves of the network characteristics in different frequency bands. Subsequently, addressing the limitations of existing regression approaches for whole brain network analysis, this paper proposed an interpretable encoder-decoder framework, Generative Age-dependent brain Network nORmative Model (GANORM). Building upon this framework, we established an age-dependent normative trajectory of the complete brain network for the entire lifespan. Finally, we validated the effectiveness of the norm using EEG datasets from multiple sites. Subsequently, we evaluated the effectiveness of GANORM, and the tested performances of BPNN showed the R^2 was 0.796, the MAE was 0.081, and the RMSE was 0.013. Following established lifespan brain network norm, GANORM also exhibited good results upon verification using healthy and disease data from various sites. The deviation scores from the normative mean for the healthy control group were significantly smaller than those of the disease group.

cs.NE↗

arXiv2025

Factorized Tail Volatility Model: Augmenting Excess-over-Threshold Method for High-Dimensional Hevay-Tailed Data

Ecess-over-Threshold method is a crucial technique in extreme value analysis, which approximately models larger observations over a threshold using a Generalized Pareto Distribution. This paper presents a comprehensive framework for analyzing tail risk in high-dimensional data by introducing the Factorized Tail Volatility Model (FTVM) and integrating it with central quantile models through the EoT method. This integrated framework is termed the FTVM-EoT method. In this framework, a quantile-related high-dimensional data model is employed to select an appropriate threshold at the central quantile for the EoT method, while the FTVM captures heteroscedastic tail volatility by decomposing tail quantiles into a low-rank linear factor structure and a heavy-tailed idiosyncratic component. The FTVM-EoT method is highly flexible, allowing for the joint modeling of central, intermediate, and extreme quantiles of high-dimensional data, thereby providing a holistic approach to tail risk analysis. In addition, we develop an iterative estimation algorithm for the FTVM-EoT method and establish the asymptotic properties of the estimators for latent factors, loadings, intermediate quantiles, and extreme quantiles. A validation procedure is introduced, and an information criterion is proposed for optimal factor selection. Simulation studies demonstrate that the FTVM-EoT method consistently outperforms existing methods at intermediate and extreme quantiles.

stat.ME↗

arXiv2025

TimeFilter: Patch-Specific Spatial-Temporal Graph Filtration for Time Series Forecasting

Time series forecasting methods generally fall into two main categories: Channel Independent (CI) and Channel Dependent (CD) strategies. While CI overlooks important covariate relationships, CD captures all dependencies without distinction, introducing noise and reducing generalization. Recent advances in Channel Clustering (CC) aim to refine dependency modeling by grouping channels with similar characteristics and applying tailored modeling techniques. However, coarse-grained clustering struggles to capture complex, time-varying interactions effectively. To address these challenges, we propose TimeFilter, a GNN-based framework for adaptive and fine-grained dependency modeling. After constructing the graph from the input sequence, TimeFilter refines the learned spatial-temporal dependencies by filtering out irrelevant correlations while preserving the most critical ones in a patch-specific manner. Extensive experiments on 13 real-world datasets from diverse application domains demonstrate the state-of-the-art performance of TimeFilter. The code is available at https://github.com/TROUBADOUR000/TimeFilter.

cs.LG↗

arXiv2025

Unraveling Interwoven Roles of Large Language Models in Authorship Privacy: Obfuscation, Mimicking, and Verification

Recent advancements in large language models (LLMs) have been fueled by large scale training corpora drawn from diverse sources such as websites, news articles, and books. These datasets often contain explicit user information, such as person names and addresses, that LLMs may unintentionally reproduce in their generated outputs. Beyond such explicit content, LLMs can also leak identity revealing cues through implicit signals such as distinctive writing styles, raising significant concerns about authorship privacy. There are three major automated tasks in authorship privacy, namely authorship obfuscation (AO), authorship mimicking (AM), and authorship verification (AV). Prior research has studied AO, AM, and AV independently. However, their interplays remain under explored, which leaves a major research gap, especially in the era of LLMs, where they are profoundly shaping how we curate and share user generated content, and the distinction between machine generated and human authored text is also increasingly blurred. This work then presents the first unified framework for analyzing the dynamic relationships among LLM enabled AO, AM, and AV in the context of authorship privacy. We quantify how they interact with each other to transform human authored text, examining effects at a single point in time and iteratively over time. We also examine the role of demographic metadata, such as gender, academic background, in modulating their performances, inter-task dynamics, and privacy risks. All source code will be publicly available.

cs.CL↗

arXiv2025

Chain-Talker: Chain Understanding and Rendering for Empathetic Conversational Speech Synthesis

Conversational Speech Synthesis (CSS) aims to align synthesized speech with the emotional and stylistic context of user-agent interactions to achieve empathy. Current generative CSS models face interpretability limitations due to insufficient emotional perception and redundant discrete speech coding. To address the above issues, we present Chain-Talker, a three-stage framework mimicking human cognition: Emotion Understanding derives context-aware emotion descriptors from dialogue history; Semantic Understanding generates compact semantic codes via serialized prediction; and Empathetic Rendering synthesizes expressive speech by integrating both components. To support emotion modeling, we develop CSS-EmCap, an LLM-driven automated pipeline for generating precise conversational speech emotion captions. Experiments on three benchmark datasets demonstrate that Chain-Talker produces more expressive and empathetic speech than existing methods, with CSS-EmCap contributing to reliable emotion modeling. The code and demos are available at: https://github.com/AI-S2-Lab/Chain-Talker.

cs.SD↗

arXiv2025

TimeBridge: Non-Stationarity Matters for Long-term Time Series Forecasting

Non-stationarity poses significant challenges for multivariate time series forecasting due to the inherent short-term fluctuations and long-term trends that can lead to spurious regressions or obscure essential long-term relationships. Most existing methods either eliminate or retain non-stationarity without adequately addressing its distinct impacts on short-term and long-term modeling. Eliminating non-stationarity is essential for avoiding spurious regressions and capturing local dependencies in short-term modeling, while preserving it is crucial for revealing long-term cointegration across variates. In this paper, we propose TimeBridge, a novel framework designed to bridge the gap between non-stationarity and dependency modeling in long-term time series forecasting. By segmenting input series into smaller patches, TimeBridge applies Integrated Attention to mitigate short-term non-stationarity and capture stable dependencies within each variate, while Cointegrated Attention preserves non-stationarity to model long-term cointegration across variates. Extensive experiments show that TimeBridge consistently achieves state-of-the-art performance in both short-term and long-term forecasting. Additionally, TimeBridge demonstrates exceptional performance in financial forecasting on the CSI 500 and S&P 500 indices, further validating its robustness and effectiveness. Code is available at https://github.com/Hank0626/TimeBridge.

cs.LG↗

arXiv2025

LSR-IGRU: Stock Trend Prediction Based on Long Short-Term Relationships and Improved GRU

Stock price prediction is a challenging problem in the field of finance and receives widespread attention. In recent years, with the rapid development of technologies such as deep learning and graph neural networks, more research methods have begun to focus on exploring the interrelationships between stocks. However, existing methods mostly focus on the short-term dynamic relationships of stocks and directly integrating relationship information with temporal information. They often overlook the complex nonlinear dynamic characteristics and potential higher-order interaction relationships among stocks in the stock market. Therefore, we propose a stock price trend prediction model named LSR-IGRU in this paper, which is based on long short-term stock relationships and an improved GRU input. Firstly, we construct a long short-term relationship matrix between stocks, where secondary industry information is employed for the first time to capture long-term relationships of stocks, and overnight price information is utilized to establish short-term relationships. Next, we improve the inputs of the GRU model at each step, enabling the model to more effectively integrate temporal information and long short-term relationship information, thereby significantly improving the accuracy of predicting stock trend changes. Finally, through extensive experiments on multiple datasets from stock markets in China and the United States, we validate the superiority of the proposed LSR-IGRU model over the current state-of-the-art baseline models. We also apply the proposed model to the algorithmic trading system of a financial company, achieving significantly higher cumulative portfolio returns compared to other baseline methods. Our sources are released at https://github.com/ZP1481616577/Baselines_LSR-IGRU.

q-fin.ST↗

arXiv2025

Reinforcement Learning with Continuous Actions Under Unmeasured Confounding

This paper addresses the challenge of offline policy learning in reinforcement learning with continuous action spaces when unmeasured confounders are present. While most existing research focuses on policy evaluation within partially observable Markov decision processes (POMDPs) and assumes discrete action spaces, we advance this field by establishing a novel identification result to enable the nonparametric estimation of policy value for a given target policy under an infinite-horizon framework. Leveraging this identification, we develop a minimax estimator and introduce a policy-gradient-based algorithm to identify the in-class optimal policy that maximizes the estimated policy value. Furthermore, we provide theoretical results regarding the consistency, finite-sample error bound, and regret bound of the resulting optimal policy. Extensive simulations and a real-world application using the German Family Panel data demonstrate the effectiveness of our proposed methodology.

stat.ML↗

arXiv2025

Adaptive Multi-Scale Decomposition Framework for Time Series Forecasting

Transformer-based and MLP-based methods have emerged as leading approaches in time series forecasting (TSF). While Transformer-based methods excel in capturing long-range dependencies, they suffer from high computational complexities and tend to overfit. Conversely, MLP-based methods offer computational efficiency and adeptness in modeling temporal dynamics, but they struggle with capturing complex temporal patterns effectively. To address these challenges, we propose a novel MLP-based Adaptive Multi-Scale Decomposition (AMD) framework for TSF. Our framework decomposes time series into distinct temporal patterns at multiple scales, leveraging the Multi-Scale Decomposable Mixing (MDM) block to dissect and aggregate these patterns in a residual manner. Complemented by the Dual Dependency Interaction (DDI) block and the Adaptive Multi-predictor Synthesis (AMS) block, our approach effectively models both temporal and channel dependencies and utilizes autocorrelation to refine multi-scale data integration. Comprehensive experiments demonstrate that our AMD framework not only overcomes the limitations of existing methods but also consistently achieves state-of-the-art performance in both long-term and short-term forecasting tasks across various datasets, showcasing superior efficiency. Code is available at https://github.com/TROUBADOUR000/AMD

cs.LG↗

arXiv2025

Global Group Fairness in Federated Learning via Function Tracking

We investigate group fairness regularizers in federated learning, aiming to train a globally fair model in a distributed setting. Ensuring global fairness in distributed training presents unique challenges, as fairness regularizers typically involve probability metrics between distributions across all clients and are not naturally separable by client. To address this, we introduce a function-tracking scheme for the global fairness regularizer based on a Maximum Mean Discrepancy (MMD), which incurs a small communication overhead. This scheme seamlessly integrates into most federated learning algorithms while preserving rigorous convergence guarantees, as demonstrated in the context of FedAvg. Additionally, when enforcing differential privacy, the kernel-based MMD regularization enables straightforward analysis through a change of kernel, leveraging an intuitive interpretation of kernel convolution. Numerical experiments confirm our theoretical insights.

cs.LG↗

arXiv2025

Bootstrap-based Inference for Bivariate Heteroscedastic Extremes with a Changing Tail Copula

This paper introduces a copula-based model for independent but non-identically distributed data with heteroscedastic extremes marginal and changing tail dependence structures. We establish a unified framework for inference by proving the weak convergence of the bivariate sequential tail empirical process and its empirical bootstrap counterpart. We derive the asymptotic properties of several estimators on the tail, including the quasi-tail copula, integrated scedasis function, and Hill estimator, treating them as functionals of the bivariate sequential tail empirical process. This process-centric approach enables the development of bootstrap-based methods and ensures the theoretical validity of the derived statistics. As an application of our inference method, we propose bootstrap-based tests for the equivalence of extreme value indices, the equivalence of scedasis functions, and non-changing tail dependence when marginal scedasis functions are identical. Our simulations validate the robustness and efficiency of the bootstrap-based tests.

stat.ME↗

arXiv2025

Gravitational waves from metastable cosmic strings in the delayed scaling scenario

Recent observations by pulsar timing arrays (PTAs) such as NANOGrav, EPTA, PPTA, and CPTA suggest the presence of nanohertz stochastic gravitational wave background (GWB). While such signals could be explained by gravitational waves from a network of metastable cosmic strings (CSs), standard scenarios involving the Kibble-Zurek mechanism triggered by a thermal potential face significant challenges. Specifically, these scenarios predict a GWB spectrum inconsistent with the non-detection at higher frequencies by LIGO-Virgo-KAGRA (LVK) for CSs with relatively large string tension. It is also difficult to prevent the monopole forming phase transition just before the CS forming symmetry breaking, which spoils the CS network formation. In contrast, a delayed scaling scenario, where the CSs start to emit GWs at a later time due to the dilution during inflation, alleviates these issues. This scenario allows for a larger string tension while monopoles are sufficiently diluted such that the CS network safely forms. In this study, we clarify the spectrum of stochastic GWB from metastable CSs in the delayed scaling scenario, consistent with the PTA observations while satisfying the LVK constraints. Furthermore, we explore its potential signatures at frequencies accessible to other detectors such as LVK as well as LISA, Taiji, and TianQin or DECIGO and BBO. We also discuss the implications on inflation and underlying UV theories, such as the grand unified theories.

astro-ph.CO↗

arXiv2025

Understanding the Functional Roles of Modelling Components in Spiking Neural Networks

Spiking neural networks (SNNs), inspired by the neural circuits of the brain, are promising in achieving high computational efficiency with biological fidelity. Nevertheless, it is quite difficult to optimize SNNs because the functional roles of their modelling components remain unclear. By designing and evaluating several variants of the classic model, we systematically investigate the functional roles of key modelling components, leakage, reset, and recurrence, in leaky integrate-and-fire (LIF) based SNNs. Through extensive experiments, we demonstrate how these components influence the accuracy, generalization, and robustness of SNNs. Specifically, we find that the leakage plays a crucial role in balancing memory retention and robustness, the reset mechanism is essential for uninterrupted temporal processing and computational efficiency, and the recurrence enriches the capability to model complex dynamics at a cost of robustness degradation. With these interesting observations, we provide optimization suggestions for enhancing the performance of SNNs in different scenarios. This work deepens the understanding of how SNNs work, which offers valuable guidance for the development of more effective and robust neuromorphic models.

cs.NE↗

arXiv2025

Direct Expression for One-Loop Tensor Reduction with Lorentz Indices via Generating Function

In recent work, we derived a direct expression for one-loop tensor reduction using generating functions and Feynman parametrization in projective space, avoiding recursive relations. However, for practical applications, this expression still presents two challenges: (1) While the final reduction coefficients are expressed in terms of the dimension D and Mandelstam variables, the given expression explicitly contains irrational functions; (2) The expression involves an auxiliary vector R, which can be eliminated via differentiation $\frac{\partial}{\partial R}$, but the presence of irrational terms making differentiation cumbersome. (3) Most practical applications require the tensor form with Lorentz indices. In this paper, we provide a rational form of the reduction coefficients with Lorentz indices, free from recursion. Additionally, We provide a pure Wolfram Mathematica implementation of the code. Our practical tests demonstrate that this direct expression achieves significantly higher computational efficiency compared to the traditional Passarino-Veltman (PV) reduction or other recursion-based methods.

hep-th↗

arXiv2025

Multi-modal and Multi-scale Spatial Environment Understanding for Immersive Visual Text-to-Speech

Visual Text-to-Speech (VTTS) aims to take the environmental image as the prompt to synthesize the reverberant speech for the spoken content. The challenge of this task lies in understanding the spatial environment from the image. Many attempts have been made to extract global spatial visual information from the RGB space of an spatial image. However, local and depth image information are crucial for understanding the spatial environment, which previous works have ignored. To address the issues, we propose a novel multi-modal and multi-scale spatial environment understanding scheme to achieve immersive VTTS, termed M2SE-VTTS. The multi-modal aims to take both the RGB and Depth spaces of the spatial image to learn more comprehensive spatial information, and the multi-scale seeks to model the local and global spatial knowledge simultaneously. Specifically, we first split the RGB and Depth images into patches and adopt the Gemini-generated environment captions to guide the local spatial understanding. After that, the multi-modal and multi-scale features are integrated by the local-aware global spatial understanding. In this way, M2SE-VTTS effectively models the interactions between local and global spatial contexts in the multi-modal spatial environment. Objective and subjective evaluations suggest that our model outperforms the advanced baselines in environmental speech generation. The code and audio samples are available at: https://github.com/AI-S2-Lab/M2SE-VTTS.

cs.CV↗

arXiv2025

Few-Shot Adaptation of Training-Free Foundation Model for 3D Medical Image Segmentation

Vision foundation models have achieved remarkable progress across various image analysis tasks. In the image segmentation task, foundation models like the Segment Anything Model (SAM) enable generalizable zero-shot segmentation through user-provided prompts. However, SAM primarily trained on natural images, lacks the domain-specific expertise of medical imaging. This limitation poses challenges when applying SAM to medical image segmentation, including the need for extensive fine-tuning on specialized medical datasets and a dependency on manual prompts, which are both labor-intensive and require intervention from medical experts. This work introduces the Few-shot Adaptation of Training-frEe SAM (FATE-SAM), a novel method designed to adapt the advanced Segment Anything Model 2 (SAM2) for 3D medical image segmentation. FATE-SAM reassembles pre-trained modules of SAM2 to enable few-shot adaptation, leveraging a small number of support examples to capture anatomical knowledge and perform prompt-free segmentation, without requiring model fine-tuning. To handle the volumetric nature of medical images, we incorporate a Volumetric Consistency mechanism that enhances spatial coherence across 3D slices. We evaluate FATE-SAM on multiple medical imaging datasets and compare it with supervised learning methods, zero-shot SAM approaches, and fine-tuned medical SAM methods. Results show that FATE-SAM delivers robust and accurate segmentation while eliminating the need for large annotated datasets and expert intervention. FATE-SAM provides a practical, efficient solution for medical image segmentation, making it more accessible for clinical applications.

cs.CV↗

arXiv2024

Contextual Bilevel Reinforcement Learning for Incentive Alignment

The optimal policy in various real-world strategic decision-making problems depends both on the environmental configuration and exogenous events. For these settings, we introduce Contextual Bilevel Reinforcement Learning (CB-RL), a stochastic bilevel decision-making model, where the lower level consists of solving a contextual Markov Decision Process (CMDP). CB-RL can be viewed as a Stackelberg Game where the leader and a random context beyond the leader's control together decide the setup of many MDPs that potentially multiple followers best respond to. This framework extends beyond traditional bilevel optimization and finds relevance in diverse fields such as RLHF, tax design, reward shaping, contract theory and mechanism design. We propose a stochastic Hyper Policy Gradient Descent (HPGD) algorithm to solve CB-RL, and demonstrate its convergence. Notably, HPGD uses stochastic hypergradient estimates, based on observations of the followers' trajectories. Therefore, it allows followers to use any training procedure and the leader to be agnostic of the specific algorithm, which aligns with various real-world scenarios. We further consider the setting when the leader can influence the training of followers and propose an accelerated algorithm. We empirically demonstrate the performance of our algorithm for reward shaping and tax design.

math.OC↗