Searcharxiv⌕ Search

arXiv subjects

Di Yang

Publications and source records attributed to Di Yang.

At least 55 records · Page 3Linked to original sources

On Gromov--Witten invariants of $\mathbb{P}^1$-orbifolds and topological difference equations

Let $(m_1, m_2)$ be a pair of positive integers. Denote by $\mathbb{P}^1$ the complex projective line, and by $\mathbb{P}^1_{m_1,m_2}$ the orbifold complex projective line obtained from $\mathbb{P}^1$ by adding $\mathbb{Z}_{m_1}$ and $\mathbb{Z}_{m_2}$ orbifold points. In this paper we introduce a matrix linear difference equation, prove existence and uniqueness of its formal Puiseux-series solutions, and use them to give conjectural formulas for $k$-point ($k\ge2$) functions of Gromov--Witten invariants of $\mathbb{P}^1_{m_1,m_2}$. Explicit expressions of the unique solutions are also obtained. We carry out concrete computations of the first few invariants by using the conjectural formulas. For the case when one of $m_1, m_2$ equals 1, we prove validity of the conjectural formulas.

math-ph↗

Uncovering Disparities in Rideshare Drivers Earning and Work Patterns: A Case Study of Chicago

Ride-sharing services are revolutionizing urban mobility while simultaneously raising significant concerns regarding fairness and driver equity. This study employs Chicago Trip Network Provider dataset to investigate disparities in ride-sharing earnings between 2018 and 2023. Our analysis reveals marked temporal shifts, including an earnings surge in early 2021 followed by fluctuations and a decline in inflation-adjusted income, as well as pronounced spatial disparities, with drivers in Central and airport regions earning substantially more than those in peripheral areas. Recognizing the limitations of trip-level data, we introduce a novel trip-driver assignment algorithm to reconstruct plausible daily work patterns, uncovering distinct driver clusters with varied earning profiles. Notably, drivers operating during late-evening and overnight hours secure higher per-trip and hourly rates, while emerging groups in low-demand regions face significant earnings deficits. Our findings call for more transparent pricing models and a re-examination of platform design to promote equitable driver outcomes.

cs.HC↗

Combinatorics and large genus asymptotics of the Brézin--Gross--Witten numbers

In this paper, we study combinatorial and asymptotic properties of some interesting rational numbers called the Brézin--Gross--Witten (BGW) numbers, which can be represented as the intersection numbers of psi and Theta classes on the moduli space of stable algebraic curves. In particular, we discover and prove the uniform large genus leading asymptotics of certain normalized BGW numbers, and give a new proof of the polynomiality phenomenon for the large genus asymptotics. We also propose, with extensive numerical data, several new conjectures including monotonicity and integrality on the BGW numbers. Applications to the Painlevé II hierarchy and to the BGW-kappa numbers are given.

math-ph↗

RoVRM: A Robust Visual Reward Model Optimized via Auxiliary Textual Preference Data

Large vision-language models (LVLMs) often fail to align with human preferences, leading to issues like generating misleading content without proper visual context (also known as hallucination). A promising solution to this problem is using human-preference alignment techniques, such as best-of-n sampling and reinforcement learning. However, these techniques face the difficulty arising from the scarcity of visual preference data, which is required to train a visual reward model (VRM). In this work, we continue the line of research. We present a Robust Visual Reward Model (RoVRM) which improves human-preference alignment for LVLMs. RoVRM leverages auxiliary textual preference data through a three-phase progressive training and optimal transport-based preference data selection to effectively mitigate the scarcity of visual preference data. We experiment with RoVRM on the commonly used vision-language tasks based on the LLaVA-1.5-7B and -13B models. Experimental results demonstrate that RoVRM consistently outperforms traditional VRMs. Furthermore, our three-phase progressive training and preference data selection approaches can yield consistent performance gains over ranking-based alignment techniques, such as direct preference optimization.

cs.CV↗

AutoCBT: An Autonomous Multi-agent Framework for Cognitive Behavioral Therapy in Psychological Counseling

Traditional in-person psychological counseling remains primarily niche, often chosen by individuals with psychological issues, while online automated counseling offers a potential solution for those hesitant to seek help due to feelings of shame. Cognitive Behavioral Therapy (CBT) is an essential and widely used approach in psychological counseling. The advent of large language models (LLMs) and agent technology enables automatic CBT diagnosis and treatment. However, current LLM-based CBT systems use agents with a fixed structure, limiting their self-optimization capabilities, or providing hollow, unhelpful suggestions due to redundant response patterns. In this work, we utilize Quora-like and YiXinLi single-round consultation models to build a general agent framework that generates high-quality responses for single-turn psychological consultation scenarios. We use a bilingual dataset to evaluate the quality of single-response consultations generated by each framework. Then, we incorporate dynamic routing and supervisory mechanisms inspired by real psychological counseling to construct a CBT-oriented autonomous multi-agent framework, demonstrating its general applicability. Experimental results indicate that AutoCBT can provide higher-quality automated psychological counseling services.

cs.CL↗

Topological Separation of Vortices

Vortices and their analysis play a critical role in the understanding of complex phenomena in turbulent flows. Traditional vortex extraction methods, notably region-based techniques, often overlook the entanglement phenomenon, resulting in the inclusion of multiple vortices within a single extracted region. Their separation is necessary for quantifying different types of vortices and their statistics. In this study, we propose a novel vortex separation method that extends the conventional contour tree-based segmentation approach with an additional step termed "layering". Upon extracting a vortical region using specified vortex criteria (e.g., $λ_2$), we initially establish topological segmentation based on the contour tree, followed by the layering process to allocate appropriate segmentation IDs to unsegmented cells, thus separating individual vortices within the region. However, these regions may still suffer from inaccurate splits, which we address statistically by leveraging the continuity of vorticity lines across the split boundaries. Our findings demonstrate a significant improvement in both the separation of vortices and the mitigation of inaccurate splits compared to prior methods.

physics.flu-dyn↗

Small Language Model as Data Prospector for Large Language Model

The quality of instruction data directly affects the performance of fine-tuned Large Language Models (LLMs). Previously, \cite{li2023one} proposed \texttt{NUGGETS}, which identifies and selects high-quality quality data from a large dataset by identifying those individual instruction examples that can significantly improve the performance of different tasks after being learnt as one-shot instances. In this work, we propose \texttt{SuperNUGGETS}, an improved variant of \texttt{NUGGETS} optimised for efficiency and performance. Our \texttt{SuperNUGGETS} uses a small language model (SLM) instead of a large language model (LLM) to filter the data for outstanding one-shot instances and refines the predefined set of tests. The experimental results show that the performance of \texttt{SuperNUGGETS} only decreases by 1-2% compared to \texttt{NUGGETS}, but the efficiency can be increased by a factor of 58. Compared to the original \texttt{NUGGETS}, our \texttt{SuperNUGGETS} has a higher utility value due to the significantly lower resource consumption.

cs.CL↗

CPsyExam: A Chinese Benchmark for Evaluating Psychology using Examinations

In this paper, we introduce a novel psychological benchmark, CPsyExam, constructed from questions sourced from Chinese language examinations. CPsyExam is designed to prioritize psychological knowledge and case analysis separately, recognizing the significance of applying psychological knowledge to real-world scenarios. From the pool of 22k questions, we utilize 4k to create the benchmark that offers balanced coverage of subjects and incorporates a diverse range of case analysis techniques.Furthermore, we evaluate a range of existing large language models~(LLMs), spanning from open-sourced to API-based models. Our experiments and analysis demonstrate that CPsyExam serves as an effective benchmark for enhancing the understanding of psychology within LLMs and enables the comparison of LLMs across various granularities.

cs.CL↗

Remarks on intersection numbers and integrable hierarchies. II. Tau-structure

For systems of evolutionary partial differential equations the tau-structure is an important notion which originated from the deep relation between integrable systems and quantum field theories. We show that, under a certain non-degeneracy condition, existence of a tau-structure implies integrability. As an example, we apply this principle to provide a new proof of the integrability of the Drinfeld--Sokolov hierarchy associated to an arbitrary Kac--Moody algebra and a choice of a vertex of its Dynkin diagram.

math-ph↗

Are Visual-Language Models Effective in Action Recognition? A Comparative Study

Current vision-language foundation models, such as CLIP, have recently shown significant improvement in performance across various downstream tasks. However, whether such foundation models significantly improve more complex fine-grained action recognition tasks is still an open question. To answer this question and better find out the future research direction on human behavior analysis in-the-wild, this paper provides a large-scale study and insight on current state-of-the-art vision foundation models by comparing their transfer ability onto zero-shot and frame-wise action recognition tasks. Extensive experiments are conducted on recent fine-grained, human-centric action recognition datasets (e.g., Toyota Smarthome, Penn Action, UAV-Human, TSU, Charades) including action classification and segmentation.

cs.CV↗

Mapping partition functions

We introduce an infinite group action on partition functions of WK type, meaning of the type of the partition function $Z^{\rm WK}$ in the famous result of Witten and Kontsevich expressing the partition function of $ψ$-class integrals on the compactified moduli space $\overline{\mathcal{M}}_{g,n}$ as a $τ$-function for the Korteweg--de Vries hierarchy. Specifically, the group which acts is the group $\mathcal{G}$ of formal power series of one variable $φ(V)=V+O(V^2)$, with group law given by composition, acting in a suitable way on the infinite tuple of variables of the partition functions. In particular, any $φ\in \mathcal{G}$ sends the Witten--Kontsevich (WK) partition function $Z^{\rm WK}$ to a new partition function $Z^φ$, which we call the WK mapping partition function associated to $φ$. We show that the genus zero part of $\log Z^φ$ is independent of $φ$ and give an explicit recursive description for its higher genus parts (loop equation), and as applications of this obtain relationships of the $ψ$-class integrals to Gaussian Unitary Ensemble and generalized Brézin--Gross--Witten correlators. In a different direction, we use $Z^φ$ to construct a new integrable hierarchy, which we call the WK mapping hierarchy associated to $φ$. We show that this hierarchy is a bihamiltonian perturbation of the Riemann--Hopf hierarchy possessing a $τ$-structure, and prove that it is a universal object for all such perturbations. Similarly, for any $φ\in\mathcal{G}$, we define the Hodge mapping partition function associated to $φ$, prove that it is integrable, and study its role in hamiltonian perturbations of the Riemann--Hopf hierarchy possessing a $τ$-structure. Finally, we establish a generalized Hodge--WK correspondence relating different Hodge mapping partition functions.

math-ph↗

The loop equation for the Burgers--KdV hierarchy

The Burgers--KdV hierarchy was introduced towards understanding intersection numbers on the moduli space of Riemann surfaces with boundaries. The goal of this paper is to derive the Dubrovin--Zhang type loop equation for the topological tau-function of the Burgers--KdV hierarchy. As as application, we provide some relations for open intersection numbers.

math-ph↗

GUE via Frobenius Manifolds. II. Loop Equations

A theorem of Dubrovin establishes the relationship between the GUE partition function and the partition function of Gromov-Witten invariants of the complex projective line. Based on this theorem we derive loop equations for the Gaussian Unitary Ensemble (GUE) partition function. We show that the GUE partition function is equal to part of the topological partition function of the non-linear Schrödinger (NLS) Frobenius manifold.

math-ph↗

Analytic theory of Legendre-type transformations for a Frobenius manifold

Let $M$ be an $n$-dimensional Frobenius manifold. Fix $κ\in\{1,\dots,n\}$. Assuming certain invertibility, Dubrovin introduced the Legendre-type transformation $S_κ$, which transforms $M$ to an $n$-dimensional Frobenius manifold $S_κ(M)$. In this paper, we show that these $S_κ(M)$ share the same monodromy data at the Fuchsian singular point of the Dubrovin connection, and that for the case when $M$ is semisimple they also share the same Stokes matrix and the same central connection matrix. A straightforward application of the monodromy identification is the following: if we know the monodromy data of some semisimple Frobenius manifold $M$, we immediately obtain those of its Legendre-type transformations. Another application gives the identification between the $κ$th partition function of a semisimple Frobenius manifold $M$ and the topological partition function of $S_κ(M)$.

math.DG↗

CPsyCoun: A Report-based Multi-turn Dialogue Reconstruction and Evaluation Framework for Chinese Psychological Counseling

Using large language models (LLMs) to assist psychological counseling is a significant but challenging task at present. Attempts have been made on improving empathetic conversations or acting as effective assistants in the treatment with LLMs. However, the existing datasets lack consulting knowledge, resulting in LLMs lacking professional consulting competence. Moreover, how to automatically evaluate multi-turn dialogues within the counseling process remains an understudied area. To bridge the gap, we propose CPsyCoun, a report-based multi-turn dialogue reconstruction and evaluation framework for Chinese psychological counseling. To fully exploit psychological counseling reports, a two-phase approach is devised to construct high-quality dialogues while a comprehensive evaluation benchmark is developed for the effective automatic evaluation of multi-turn psychological consultations. Competitive experimental results demonstrate the effectiveness of our proposed framework in psychological counseling. We open-source the datasets and model for future research at https://github.com/CAS-SIAT-XinHai/CPsyCoun

cs.CL↗

Text Modality Oriented Image Feature Extraction for Detecting Diffusion-based DeepFake

The widespread use of diffusion methods enables the creation of highly realistic images on demand, thereby posing significant risks to the integrity and safety of online information and highlighting the necessity of DeepFake detection. Our analysis of features extracted by traditional image encoders reveals that both low-level and high-level features offer distinct advantages in identifying DeepFake images produced by various diffusion methods. Inspired by this finding, we aim to develop an effective representation that captures both low-level and high-level features to detect diffusion-based DeepFakes. To address the problem, we propose a text modality-oriented feature extraction method, termed TOFE. Specifically, for a given target image, the representation we discovered is a corresponding text embedding that can guide the generation of the target image with a specific text-to-image model. Experiments conducted across ten diffusion types demonstrate the efficacy of our proposed method.

cs.CV↗

A Novel Paradigm in Solving Multiscale Problems

Multiscale phenomena manifest across various scientific domains, presenting a ubiquitous challenge in accurately and effectively simulating multiscale dynamics in complex systems. In this paper, a novel decoupling solving paradigm is proposed through modelling large-scale dynamics independently and treating small-scale dynamics as a slaved system. A Spectral Physics-informed Neural Network (PINN) is developed to characterize the small-scale system in an efficient and accurate way, addressing the challenges posed by the representation of multiscale dynamics in neural networks. The effectiveness of the method is demonstrated through extensive numerical experiments, including one-dimensional Kuramot-Sivashinsky equation, two- and three-dimensional Navier-Stokes equations, showcasing its versatility in addressing problems of fluid dynamics. Furthermore, we also delve into the application of the proposed approach to more complex problems, including non-uniform meshes, complex geometries, large-scale data with noise, and high-dimensional small-scale dynamics. The discussions about these scenarios contribute to a comprehensive understanding of the method's capabilities and limitations. By enabling the acquisition of large-scale data with minimal computational demands, coupled with the efficient and accurate characterization of small-scale dynamics via Spectral PINN, our approach offers a valuable and promising approach for researchers seeking to tackle multiscale phenomena effectively.

physics.flu-dyn↗