arXiv subjects

Xiaoqian Sun

Publications and source records attributed to Xiaoqian Sun.

15 recordsLinked to original sources

arXiv2026

R$^2$PO: Decoupling Rollout and Inference Policies for LLM Reasoning

Existing reinforcement learning methods for LLM reasoning implicitly assume that the policy generating training trajectories should coincide with the one producing inference responses. We argue that this is a misleading inductive bias: the optimization-optimal trajectory distribution favors informative gradients, whereas the inference-optimal response distribution emphasizes accuracy and consistency. Forcing both into a single policy entangles their gradients and suppresses exploration. We propose R$^2$PO (Residual Rollout Policy Optimization), which attaches a lightweight Residual Rollout-Head atop the policy to decouple training trajectories from inference responses, diversifying rollouts during training while keeping inference generation intact. Experiments show that R$^2$PO consistently outperforms baselines, with average accuracy gains of 3.4% on MATH-500 and 1.3% on APPS, alongside more diverse rollouts and reduced length bias. Our code is available at https://github.com/RRPO-ARR/Code.

cs.LG↗

arXiv2026

ActiveMem: Distributed Active Memory for Long-Horizon LLM Reasoning

Memory is essential for enabling large language model (LLM) agents to handle long-horizon reasoning tasks. Existing memory mechanisms are largely centralized, typically organizing retrieved information and interaction history within a single model context. This design imposes a fundamental trade-off: scaling reasoning trajectories risks context overload, whereas aggressive content pruning may result in irreversible information loss. Seeking a better trade-off, we draw inspiration from human cognitive systems, especially the functional complementarity between the prefrontal cortex (executive control) and the hippocampus (memory management), suggesting that such a trade-off need not be inherent, but may instead stem from centralized memory organization. To this end, we propose ActiveMem, a heterogeneous framework that decouples agent memory from the core reasoning process. Specifically, a high-level Planner utilizes distilled semantic gists to execute reasoning, while a lightweight, distributed memory system operates in parallel to actively accumulate and consolidate these gists throughout the task. Experiments on BrowseComp-Plus and GAIA show that ActiveMem achieves state-of-the-art accuracy with significantly reduced overhead, demonstrating the effectiveness of distributed active memory for long-horizon reasoning.

cs.AI↗

arXiv2024

LOGIN: A Large Language Model Consulted Graph Neural Network Training Framework

Recent prevailing works on graph machine learning typically follow a similar methodology that involves designing advanced variants of graph neural networks (GNNs) to maintain the superior performance of GNNs on different graphs. In this paper, we aim to streamline the GNN design process and leverage the advantages of Large Language Models (LLMs) to improve the performance of GNNs on downstream tasks. We formulate a new paradigm, coined "LLMs-as-Consultants," which integrates LLMs with GNNs in an interactive manner. A framework named LOGIN (LLM Consulted GNN training) is instantiated, empowering the interactive utilization of LLMs within the GNN training process. First, we attentively craft concise prompts for spotted nodes, carrying comprehensive semantic and topological information, and serving as input to LLMs. Second, we refine GNNs by devising a complementary coping mechanism that utilizes the responses from LLMs, depending on their correctness. We empirically evaluate the effectiveness of LOGIN on node classification tasks across both homophilic and heterophilic graphs. The results illustrate that even basic GNN architectures, when employed within the proposed LLMs-as-Consultants paradigm, can achieve comparable performance to advanced GNNs with intricate designs. Our codes are available at https://github.com/QiaoYRan/LOGIN.

cs.LG↗

arXiv2023

Bridged-GNN: Knowledge Bridge Learning for Effective Knowledge Transfer

The data-hungry problem, characterized by insufficiency and low-quality of data, poses obstacles for deep learning models. Transfer learning has been a feasible way to transfer knowledge from high-quality external data of source domains to limited data of target domains, which follows a domain-level knowledge transfer to learn a shared posterior distribution. However, they are usually built on strong assumptions, e.g., the domain invariant posterior distribution, which is usually unsatisfied and may introduce noises, resulting in poor generalization ability on target domains. Inspired by Graph Neural Networks (GNNs) that aggregate information from neighboring nodes, we redefine the paradigm as learning a knowledge-enhanced posterior distribution for target domains, namely Knowledge Bridge Learning (KBL). KBL first learns the scope of knowledge transfer by constructing a Bridged-Graph that connects knowledgeable samples to each target sample and then performs sample-wise knowledge transfer via GNNs.KBL is free from strong assumptions and is robust to noises in the source data. Guided by KBL, we propose the Bridged-GNN} including an Adaptive Knowledge Retrieval module to build Bridged-Graph and a Graph Knowledge Transfer module. Comprehensive experiments on both un-relational and relational data-hungry scenarios demonstrate the significant improvements of Bridged-GNN compared with SOTA methods

cs.LG↗

arXiv2023

Predicting the Silent Majority on Graphs: Knowledge Transferable Graph Neural Network

Graphs consisting of vocal nodes ("the vocal minority") and silent nodes ("the silent majority"), namely VS-Graph, are ubiquitous in the real world. The vocal nodes tend to have abundant features and labels. In contrast, silent nodes only have incomplete features and rare labels, e.g., the description and political tendency of politicians (vocal) are abundant while not for ordinary people (silent) on the twitter's social network. Predicting the silent majority remains a crucial yet challenging problem. However, most existing message-passing based GNNs assume that all nodes belong to the same domain, without considering the missing features and distribution-shift between domains, leading to poor ability to deal with VS-Graph. To combat the above challenges, we propose Knowledge Transferable Graph Neural Network (KT-GNN), which models distribution shifts during message passing and representation learning by transferring knowledge from vocal nodes to silent nodes. Specifically, we design the domain-adapted "feature completion and message passing mechanism" for node representation learning while preserving domain difference. And a knowledge transferable classifier based on KL-divergence is followed. Comprehensive experiments on real-world scenarios (i.e., company financial risk assessment and political elections) demonstrate the superior performance of our method. Our source code has been open sourced.

cs.LG↗

arXiv2023

Company-as-Tribe: Company Financial Risk Assessment on Tribe-Style Graph with Hierarchical Graph Neural Networks

Company financial risk is ubiquitous and early risk assessment for listed companies can avoid considerable losses. Traditional methods mainly focus on the financial statements of companies and lack the complex relationships among them. However, the financial statements are often biased and lagged, making it difficult to identify risks accurately and timely. To address the challenges, we redefine the problem as \textbf{company financial risk assessment on tribe-style graph} by taking each listed company and its shareholders as a tribe and leveraging financial news to build inter-tribe connections. Such tribe-style graphs present different patterns to distinguish risky companies from normal ones. However, most nodes in the tribe-style graph lack attributes, making it difficult to directly adopt existing graph learning methods (e.g., Graph Neural Networks(GNNs)). In this paper, we propose a novel Hierarchical Graph Neural Network (TH-GNN) for Tribe-style graphs via two levels, with the first level to encode the structure pattern of the tribes with contrastive learning, and the second level to diffuse information based on the inter-tribe relations, achieving effective and efficient risk assessment. Extensive experiments on the real-world company dataset show that our method achieves significant improvements on financial risk assessment over previous competing methods. Also, the extensive ablation studies and visualization comprehensively show the effectiveness of our method.

cs.LG↗

arXiv2021

Bayesian Poisson Mortality Projections with Incomplete Data

The missing data problem pervasively exists in statistical applications. Even as simple as the count data in mortality projections, it may not be available for certain age-and-year groups due to the budget limitations or difficulties in tracing research units, resulting in the follow-up estimation and prediction inaccuracies. To circumvent this data-driven challenge, we extend the Poisson log-normal Lee-Carter model to accommodate a more flexible time structure, and develop the new sampling algorithm that improves the MCMC convergence when dealing with incomplete mortality data. Via the overdispersion term and Gibbs sampler, the extended model can be re-written as the dynamic linear model so that both Kalman and sequential Kalman filters can be incorporated into the sampling scheme. Additionally, our meticulous prior settings can avoid the re-scaling step in each MCMC iteration, and allow model selection simultaneously conducted with estimation and prediction. The proposed method is applied to the mortality data of Chinese males during the period 1995-2016 to yield mortality rate forecasts for 2017-2039. The results are comparable to those based on the imputed data set, suggesting that our approach could handle incomplete data well.

stat.ME↗

arXiv2021

A Bayesian Spatial Modeling Approach to Mortality Forecasting

This paper extends Bayesian mortality projection models for multiple populations considering the stochastic structure and the effect of spatial autocorrelation among the observations. We explain high levels of overdispersion according to adjacent locations based on the conditional autoregressive model. In an empirical study, we compare different hierarchical projection models for the analysis of geographical diversity in mortality between the Japanese counties in multiple years, according to age. By a Markov chain Monte Carlo (MCMC) computation, results have demonstrated the flexibility and predictive performance of our proposed model.

stat.ME↗

arXiv2021

Modelling Universal Order Book Dynamics in Bitcoin Market

Understanding the emergence of universal features such as the stylized facts in markets is a long-standing challenge that has drawn much attention from economists and physicists. Most existing models, such as stochastic volatility models, focus mainly on price changes, neglecting the complex trading dynamics. Recently, there are increasing studies on order books, thanks to the availability of large-scale trading datasets, aiming to understand the underlying mechanisms governing the market dynamics. In this paper, we collect order-book datasets of Bitcoin platforms across three countries over millions of users and billions of daily turnovers. We find a 1+1D field theory, govern by a set of KPZ-like stochastic equations, predicts precisely the order book dynamics observed in empirical data. Despite the microscopic difference of markets, we argue the proposed effective field theory captures the correct universality class of market dynamics. We also show that the model agrees with the existing stochastic volatility models at the long-wavelength limit.

q-fin.TR↗

arXiv2020

Bayesian Poisson Log-normal Model with Regularized Time Structure for Mortality Projection of Multi-population

The improvement of mortality projection is a pivotal topic in the diverse branches related to insurance, demography, and public policy. Motivated by the thread of Lee-Carter related models, we propose a Bayesian model to estimate and predict mortality rates for multi-population. This new model features in information borrowing among populations and properly reflecting variations of data. It also provides a solution to a long-time overlooked problem: model selection for dependence structures of population-specific time parameters. By introducing a Dirac spike function, simultaneous model selection and estimation for population-specific time effects can be achieved without much extra computation cost. We use the Japanese mortality data from Human Mortality Database to illustrate the desirable properties of our model.

stat.AP↗

arXiv2020

On the degree of synchronization between air transport connectivity and COVID-19 cases at worldwide level

The current outbreak of COVID-19 is an unprecedented event in air transportation. While it was known that air transportation plays a key role in the spread of a pandemic, it is probably the first time that global aviation contributed to the planet-wide spread of a pandemic, with over 500,000 reported casualties related to the infection. In this study, we perform an analysis of the role air transportation played and how the air transportation system was changed throughout the ongoing pandemic. The major decision makers were countries (governments) and airlines, who were involved mostly in the decision and implementation of travel bans. Here, we focus on the role of countries, by analyzing the degree of synchronization between the number of cases reported in specific countries and how/when these countries reacted concerning the air transportation operations. Our study also provides a comprehensive empirical analysis on the impact of the COVID-19 pandemic on aviation at a country-level. It is hoped that the study can lead to novel insights into the prevention and control of future waves or pandemics of other types.

physics.soc-ph↗

arXiv2018

From the difference of structures to the structure of the difference

When dealing with evolving or multi-dimensional complex systems, network theory provides with elegant ways of describing their constituting components, through respectively time-varying and multi-layer complex networks. Nevertheless, the analysis of how these components are related is still an open problem. We here propose a framework for analysing the evolution of a (complex) system, by describing the structure created by the difference between multiple networks by means of the Information Content metric. As opposed to other approaches, as for instance the use of global overlap or entropies, the proposed one allows to understand if the observed changes are due to random noise, or to structural (targeted) modifications. We validate the framework by means of sets of synthetic networks, as well as networks representing real technological, social and biological evolving systems. We further propose a way of reconstructing network correlograms, which allow to convert the system's evolution to the frequency domain.

physics.soc-ph↗

arXiv2016

Efficient and Accurate Robustness Estimation for Large Complex Networks

Robustness estimation is critical for the design and maintenance of resilient networks, one of the global challenges of the 21st century. Existing studies exploit network metrics to generate attack strategies, which simulate intentional attacks in a network, and compute a metric-induced robustness estimation. While some metrics are easy to compute, e.g. degree centrality, other, more accurate, metrics require considerable computation efforts, e.g. betweennes centrality. We propose a new algorithm for estimating the robustness of a network in sub-quadratic time, i.e., significantly faster than betweenness centrality. Experiments on real-world networks and random networks show that our algorithm estimates the robustness of networks close to or even better than betweenness centrality, while being orders of magnitudes faster. Our work contributes towards scalable, yet accurate methods for robustness estimation of large complex networks.

cs.SI↗

arXiv2013

Bayes Factor Consistency for One-way Random Effects Model

In this paper, we consider Bayesian hypothesis testing for the balanced one-way random effects model. A special choice of the prior formulation for the ratio of variance components is shown to yield an explicit closed-form Bayes factor without integral representation. Furthermore, we study the consistency issue of the resulting Bayes factor under three asymptotic scenarios: either the number of units goes to infinity, the number of observations per unit goes to infinity, or both go to infinity. Finally, the behavior of the proposed approach is illustrated by simulation studies.

stat.ME↗

arXiv2012

Bayes Factor Consistency for Unbalanced ANOVA Models

In practical situations, most experimental designs often yield unbalanced data which have different numbers of observations per unit because of cost constraints, or missing data, etc. In this paper, we consider the Bayesian approach to hypothesis testing or model selection under the one-way unbalanced fixed-effects ANOVA model. We adopt Zellner's g-prior with the beta-prime distribution for g, which results in an explicit closed-form expression of the Bayes factor without integral representation. Furthermore, we investigate the model selection consistency of the Bayes factor under three different asymptotic scenarios: either the number of units goes to infinity, the number of observations per unit goes to infinity, or both go to infinity. The results presented extend some existing ones of the Bayes factor for the balanced ANOVA models in the literature.

stat.ME↗