Author name cluster

Bing Xu

Possible papers associated with this exact author name in Arrow. This page groups case-insensitive exact name matches and is not a full identity disambiguation profile.

6 papers

1 author row

AAAI Conference 2026 Conference Paper

Long-form RewardBench: Evaluating Reward Models for Long-form Generation

Hui Huang
Yancheng He
Wei Liu
Muyun Yang
Jiaheng Liu
Kehai Chen
Bing Xu
Conghui Zhu

The widespread adoption of reinforcement learning-based alignment highlights the growing importance of reward models. Various benchmarks have been built to evaluate reward models in various domains and scenarios. However, a significant gap remains in assessing reward models for long-form generation, despite its critical role in real-world applications. To bridge this, we introduce Long-form RewardBench, the first reward modeling testbed specifically designed for long-form generation. Our benchmark encompasses five key subtasks: QA, RAG, Chat, Writing, and Reasoning. We collected instruction and preference data through a meticulously designed multi-stage data collection process, and conducted extensive experiments on 20+ mainstream reward models, including both classifiers and generative models. Our findings reveal that current models still lack long-form reward modeling capabilities. Furthermore, we designed a novel Long-form Needle-in-a-Haystack Test, which revealed a correlation between reward modeling performance and the error's position within a response, as well as the overall response length, with distinct characteristics observed between classification and generative models. Finally, we demonstrate that classifier exhibit better generalizability compared to generative models trained on the same data. As the first benchmark for long-form reward modeling, this work aims to offer a robust platform for visualizing progress in this crucial area.

PDF Details DOI

AAAI Conference 2026 Conference Paper

Lost in Benchmarks? Rethinking Large Language Model Benchmarking with Item Response Theory

Hongli Zhou
Hui Huang
Ziqing Zhao
Lvyuan Han
Huicheng Wang
Kehai Chen
Muyun Yang
Wei Bao

The evaluation of large language models (LLMs) via benchmarks is widespread, yet inconsistencies between different leaderboards and poor separability among top models raise concerns about their ability to accurately reflect authentic model capabilities. This paper provides a critical analysis of benchmark effectiveness, examining mainstream prominent LLM benchmarks using results from diverse models. We first propose Pseudo-Siamese Network for Item Response Theory (PSN-IRT), an enhanced Item Response Theory framework that incorporates a rich set of item parameters within an IRT-grounded architecture. PSN-IRT can be utilized for accurate and reliable estimations of item characteristics and model abilities. Based on PSN-IRT, we conduct extensive analysis on 11 LLM benchmarks comprising 41,871 items, revealing significant and varied shortcomings in their measurement quality. Furthermore, we demonstrate that leveraging PSN-IRT is able to construct smaller benchmarks while maintaining stronger alignment with human preference.

PDF Details DOI

YNICL Journal 2023 Journal Article

Exploring the longitudinal associations of functional network connectivity and psychiatric symptom changes in youth

Lorenza Dall'Aglio
Fernando Estévez-López
Mónica López-Vicente
Bing Xu
Oktay Agcaoglu
Elias Boroda
Kelvin O. Lim
Vince D. Calhoun

BACKGROUND: Functional connectivity has been associated with psychiatric problems, both in children and adults, but inconsistencies are present across studies. Prior research has mostly focused on small clinical samples with cross-sectional designs. METHODS: We adopted a longitudinal design with repeated assessments to investigate associations between functional network connectivity (FNC) and psychiatric problems in youth (9- to 17-year-olds, two time points) from the general population. The largest single-site study of pediatric neurodevelopment was used: Generation R (N = 3,131 with data at either time point). Psychiatric symptoms were measured with the Child Behavioral Checklist as broadband internalizing and externalizing problems, and its eight specific syndrome scales (e.g., anxious-depressed). FNC was assessed with two complementary approaches. First, static FNC (sFNC) was measured with graph theory-based metrics. Second, dynamic FNC (dFNC), where connectivity is allowed to vary over time, was summarized into 5 states that participants spent time in. Cross-lagged panel models were used to investigate the longitudinal bidirectional relationships of sFNC with internalizing and externalizing problems. Similar cross-lagged panel models were run for dFNC. RESULTS: Small longitudinal relationships between dFNC and certain syndrome scales were observed, especially for baseline syndrome scales (i.e., rule-breaking, somatic complaints, thought problems, and attention problems) predicting connectivity changes. However, no association between any of the psychiatric problems (broadband and syndrome scales) with either measure of FNC survived correction for multiple testing. CONCLUSION: We found no or very modest evidence for longitudinal associations between psychiatric problems with dynamic and static FNC in this population-based sample. Differences in findings may stem from the population drawn, study design, developmental timing, and sample sizes.

Details DOI

AAAI Conference 2023 Conference Paper

Robust Causal Graph Representation Learning against Confounding Effects

Hang Gao
Jiangmeng Li
Wenwen Qiang
Lingyu Si
Bing Xu
Changwen Zheng
Fuchun Sun

The prevailing graph neural network models have achieved significant progress in graph representation learning. However, in this paper, we uncover an ever-overlooked phenomenon: the pre-trained graph representation learning model tested with full graphs underperforms the model tested with well-pruned graphs. This observation reveals that there exist confounders in graphs, which may interfere with the model learning semantic information, and current graph representation learning methods have not eliminated their influence. To tackle this issue, we propose Robust Causal Graph Representation Learning (RCGRL) to learn robust graph representations against confounding effects. RCGRL introduces an active approach to generate instrumental variables under unconditional moment restrictions, which empowers the graph representation learning model to eliminate confounders, thereby capturing discriminative information that is causally related to downstream predictions. We offer theorems and proofs to guarantee the theoretical effectiveness of the proposed approach. Empirically, we conduct extensive experiments on a synthetic dataset and multiple benchmark datasets. Experimental results demonstrate the effectiveness and generalization ability of RCGRL. Our codes are available at https://github.com/hang53/RCGRL.

PDF Details DOI

IS Journal 2018 Journal Article

Accurate Object Segmentation for Video Sequences via Temporal-Spatial-Frequency Saliency Model

Bing Xu
Yanxiong Niu

In practical applications of video object segmentation, a major challenge is how to increase the accuracy and robustness of complex dynamic scenes and achieve positive scalability under camera displacement conditions. This paper proposes a novel approach to object segmentation by combining information in temporal, spatial, and frequency domains to realize complementary superiority. Four components, which are motion, color, luminance, and spectral residual from three domains, are applied to structure the temporal-spatial-frequency saliency (TSFS) model. A determining rule is defined to merge the components into a final model. The proposed model is evaluated on two representative video sequences datasets. The experimental results indicate that the model is more accurate, robust, and effective than other state-of-the-art methods, and can satisfy the requirements of segmenting the object in complex dynamic scenes and large camera displacement conditions.

Details DOI

NeurIPS Conference 2014 Conference Paper

Generative Adversarial Nets

Ian Goodfellow
Jean Pouget-Abadie
Mehdi Mirza
Bing Xu
David Warde-Farley
Sherjil Ozair
Aaron Courville
Yoshua Bengio

We propose a new framework for estimating generative models via adversarial nets, in which we simultaneously train two models: a generative model G that captures the data distribution, and a discriminative model D that estimates the probability that a sample came from the training data rather than G. The training procedure for G is to maximize the probability of D making a mistake. This framework corresponds to a minimax two-player game. In the space of arbitrary functions G and D, a unique solution exists, with G recovering the training data distribution and D equal to 1/2 everywhere. In the case where G and D are defined by multilayer perceptrons, the entire system can be trained with backpropagation. There is no need for any Markov chains or unrolled approximate inference networks during either training or generation of samples. Experiments demonstrate the potential of the framework through qualitative and quantitatively evaluation of the generated samples.

PDF Details