Xiaolong Jin

ICLR Conference 2025 Conference Paper

Real-Time Video Generation with Pyramid Attention Broadcast

• Xuanlei Zhao
• Xiaolong Jin
• Kai Wang 0036
• Yang You 0001

We present Pyramid Attention Broadcast (PAB), a real-time, high quality and training-free approach for DiT-based video generation. Our method is founded on the observation that attention difference in the diffusion process exhibits a U-shaped pattern, indicating significant redundancy. We mitigate this by broadcasting attention outputs to subsequent steps in a pyramid style. It applies different broadcast strategies to each attention based on their variance for best efficiency. We further introduce broadcast sequence parallel for more efficient distributed inference. PAB demonstrates up to 10.5x speedup across three models compared to baselines, achieving real-time generation for up to 720p videos. We anticipate that our simple yet effective method will serve as a robust baseline and facilitate future research and application for video generation.

NeurIPS Conference 2025 Conference Paper

SuperGPQA: Scaling LLM Evaluation across 285 Graduate Disciplines

• Xeron Du
• Yifan Yao
• Kaijing Ma
• Bingli Wang
• Tianyu Zheng
• Minghao Liu
• Yiming Liang
• Xiaolong Jin

Large language models (LLMs) have demonstrated remarkable proficiency in mainstream academic disciplines such as mathematics, physics, and computer science. However, human knowledge encompasses over 200 specialized disciplines, far exceeding the scope of existing benchmarks. The capabilities of LLMs in many of these specialized fields-particularly in light industry, agriculture, and service-oriented disciplines-remain inadequately evaluated. To address this gap, we present SuperGPQA, a comprehensive benchmark that evaluates graduate-level knowledge and reasoning capabilities across 285 disciplines. Our benchmark employs a novel Human-LLM collaborative filtering mechanism to eliminate trivial or ambiguous questions through iterative refinement based on both LLM responses and expert feedback. Our experimental results reveal significant room for improvement in the performance of current state-of-the-art LLMs across diverse knowledge domains (e. g. , the reasoning-focused model Gemini-2. 5-Pro achieved the highest accuracy of 63. 56% on SuperGPQA), highlighting the considerable gap between current model capabilities and artificial general intelligence. Additionally, we present comprehensive insights from our management of a large-scale annotation process, involving over 80 expert annotators and an interactive Human-LLM collaborative system, offering valuable methodological guidance for future research initiatives of comparable scope.

NeurIPS Conference 2024 Conference Paper

BiScope: AI-generated Text Detection by Checking Memorization of Preceding Tokens

• Hanxi Guo
• Siyuan Cheng
• Xiaolong Jin
• Zhuo Zhang
• Kaiyuan Zhang
• Guanhong Tao
• Guangyu Shen
• Xiangyu Zhang

Detecting text generated by Large Language Models (LLMs) is a pressing need in order to identify and prevent misuse of these powerful models in a wide range of applications, which have highly undesirable consequences such as misinformation and academic dishonesty. Given a piece of subject text, many existing detection methods work by measuring the difficulty of LLM predicting the next token in the text from their prefix. In this paper, we make a critical observation that how well the current token’s output logits memorizes the closely preceding input tokens also provides strong evidence. Therefore, we propose a novel bi-directional calculation method that measures the cross-entropy losses between an output logits and the ground-truth token (forward) and between the output logits and the immediately preceding input token (backward). A classifier is trained to make the final prediction based on the statistics of these losses. We evaluate our system, named BISCOPE, on texts generated by five latest commercial LLMs across five heterogeneous datasets, including both natural language and code. BISCOPE demonstrates superior detection accuracy and robustness compared to six existing baseline methods, exceeding the state-of-the-art non-commercial methods’ detection accuracy by over 0. 30 F1 score, achieving over 0. 95 detection F1 score on average. It also outperforms the best commercial tool GPTZero that is based on a commercial LLM trained with an enormous volume of data. Code is available at https: //github. com/MarkGHX/BiScope.

AAAI Conference 2023 Conference Paper

Rich Event Modeling for Script Event Prediction

• Long Bai
• Saiping Guan
• Zixuan Li
• Jiafeng Guo
• Xiaolong Jin
• Xueqi Cheng

Script is a kind of structured knowledge extracted from texts, which contains a sequence of events. Based on such knowledge, script event prediction aims to predict the subsequent event. To do so, two aspects should be considered for events, namely, event description (i.e., what the events should contain) and event encoding (i.e., how they should be encoded). Most existing methods describe an event by a verb together with a few core arguments (i.e., subject, object, and indirect object), which are not precise enough. In addition, existing event encoders are limited to a fixed number of arguments, which are not flexible enough to deal with extra information. Thus, in this paper, we propose the Rich Event Prediction (REP) framework for script event prediction. Fundamentally, it is based on the proposed rich event description, which enriches the existing ones with three kinds of important information, namely, the senses of verbs, extra semantic roles, and types of participants. REP contains an event extractor to extract such information from texts. Based on the extracted rich information, a predictor then selects the most probable subsequent event. The core component of the predictor is a transformer-based event encoder that integrates the above information flexibly. Experimental results on the widely used Gigaword Corpus show the effectiveness of the proposed framework.

AAAI Conference 2020 Short Paper

Entity Type Enhanced Neural Model for Distantly Supervised Relation Extraction (Student Abstract)

• Long Bai
• Xiaolong Jin
• Chuanzhi Zhuang
• Xueqi Cheng

Distantly Supervised Relation Extraction (DSRE) has been widely studied, since it can automatically extract relations from very large corpora. However, existing DSRE methods only use little semantic information about entities, such as the information of entity type. Thus, in this paper, we propose a method for integrating entity type information into a neural network based DSRE model. It also adopts two attention mechanisms, namely, sentence attention and type attention. The former selects the representative sentences for a sentence bag, while the latter selects appropriate type information for entities. Experimental comparison with existing methods on a benchmark dataset demonstrates its merits.

AAAI Conference 2020 Short Paper

Link Prediction between Group Entities in Knowledge Graphs (Student Abstract)

• Jialin Su
• Yuanzhuo Wang
• Xiaolong Jin
• Yantao Jia
• Xueqi Cheng

Link prediction in knowledge graphs (KGs) aims at predicting potential links between entities in KGs. Existing knowledge graph embedding (KGE) based methods represent individual entities and links in KGs as vectors in low-dimension space. However, these methods focus mainly on the link prediction of individual entities, yet neglect that between group entities, which exist widely in real-world KGs. In this paper, we propose a KGE based method, called GTransA, for link prediction between group entities in a heterogeneous network by integrating individual entity links into group entity links during prediction. Experiments show that GTransA decreases mean rank by 5. 4%, compared to TransA.

IJCAI Conference 2017 Conference Paper

Cross-Domain Recommendation: An Embedding and Mapping Approach

• Tong Man
• Huawei Shen
• Xiaolong Jin
• Xueqi Cheng

Data sparsity is one of the most challenging problems for recommender systems. One promising solution to this problem is cross-domain recommendation, i. e. , leveraging feedbacks or ratings from multiple domains to improve recommendation performance in a collective manner. In this paper, we propose an Embedding and Mapping framework for Cross-Domain Recommendation, called EMCDR. The proposed EMCDR framework distinguishes itself from existing cross-domain recommendation models in two aspects. First, a multi-layer perceptron is used to capture the nonlinear mapping function across domains, which offers high flexibility for learning domain-specific features of entities in each domain. Second, only the entities with sufficient data are used to learn the mapping function, guaranteeing its robustness to noise caused by data sparsity in single domain. Extensive experiments on two cross-domain recommendation scenarios demonstrate that EMCDR significantly outperforms state-of-the-art cross-domain recommendation methods.

AAAI Conference 2016 Conference Paper

Locally Adaptive Translation for Knowledge Graph Embedding

• Yantao Jia
• Yuanzhuo Wang
• Hailun Lin
• Xiaolong Jin
• Xueqi Cheng

Knowledge graph embedding aims to represent entities and relations in a large-scale knowledge graph as elements in a continuous vector space. Existing methods, e. g. , TransE and TransH, learn embedding representation by defining a global margin-based loss function over the data. However, the optimal loss function is determined during experiments whose parameters are examined among a closed set of candidates. Moreover, embeddings over two knowledge graphs with different entities and relations share the same set of candidate loss functions, ignoring the locality of both graphs. This leads to the limited performance of embedding related applications. In this paper, we propose a locally adaptive translation method for knowledge graph embedding, called TransA, to find the optimal loss function by adaptively determining its margin over different knowledge graphs. Experiments on two benchmark data sets demonstrate the superiority of the proposed method, as compared to the-state-of-the-art ones.

TIST Journal 2016 Journal Article

Location Prediction

• Yantao Jia
• Yuanzhuo Wang
• Xiaolong Jin
• Xueqi Cheng

In social networks, predicting a user’s location mainly depends on those of his/her friends, where the key lies in how to select his/her most influential friends. In this article, we analyze the theoretically maximal accuracy of location prediction based on friends’ locations and compare it with the practical accuracy obtained by the state-of-the-art location prediction methods. Upon observing a big gap between the theoretical and practical accuracy, we propose a new strategy for selecting influential friends in order to improve the practical location prediction accuracy. Specifically, several features are defined to measure the influence of the friends on a user’s location, based on which we put forth a sequential random-walk-with-restart procedure to rank the friends of the user in terms of their influence. By dynamically selecting the top N most influential friends of the user per time slice, we develop a temporal-spatial Bayesian model to characterize the dynamics of friends’ influence for location prediction. Finally, extensive experimental results on datasets of real social networks demonstrate that the proposed influential friend selection method and temporal-spatial Bayesian model can significantly improve the accuracy of location prediction.

IJCAI Conference 2016 Conference Paper

Predict Anchor Links across Social Networks via an Embedding Approach

• Tong Man
• Huawei Shen
• Shenghua Liu
• Xiaolong Jin
• Xueqi Cheng

Predicting anchor links across social networks has important implications to an array of applications, including cross-network information diffusion and cross-domain recommendation. One challenging problem is: whether and to what extent we can address the anchor link prediction problem, if only structural information of networks is available. Most existing methods, unsupervised or supervised, directly work on networks themselves rather than on their intrinsic structural regularities, and thus their effectiveness is sensitive to the high dimension and sparsity of networks. To offer a robust method, we propose a novel supervised model, called PALE, which employs network embedding with awareness of observed anchor links as supervised information to capture the major and specific structural regularities and further learns a stable cross-network mapping for predicting anchor links. Through extensive experiments on two realistic datasets, we demonstrate that PALE significantly outperforms the state-of-the-art methods.

EAAI Journal 2004 Journal Article

Characterizing autonomic task distribution and handling in grids

• Xiaolong Jin
• Jiming Liu

Autonomic computing and grid computing are two innovative information technologies. This paper explores autonomic computing in grids. A grid usually connects a huge number of computers over the Internet as a complex computational system, therefore there also exists a need to implement autonomic computing. Generally speaking, in grids, numerous tasks are distributed to grid nodes in a decentralized fashion. In this case, one of the important problems is how to implement autonomic task distribution and handling. To address this problem, this paper presents an agent-based autonomic task distribution and handling paradigm for grids. It then provides models for characterizing two typical scenarios based on this paradigm, i. e. , task distribution scenario in a short time interval and task handling scenario in a long time interval. Through modeling as well as case studies, some key features in the dynamics of the autonomic task distribution and handling in grids are characterized. The paper also investigates the effects of some important parameters on the autonomic task distribution and handling.

AAMAS Conference 2003 Conference Paper

Agent Network Topology and Complexity

• Xiaolong Jin
• Jiming Liu
• Yuanshi Wang
• Yi Tang

AAMAS Conference 2003 Conference Paper

Macroscopic Model of Agent-based Load Balancing on Grids

• Yuanshi Wang
• Jiming Liu
• Xiaolong Jin
• Tang Yi