Author name cluster

Hao Yuan

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.

7 papers

2 author rows

NeurIPS Conference 2025 Conference Paper

OPTFM: A Scalable Multi-View Graph Transformer for Hierarchical Pre-Training in Combinatorial Optimization

Hao Yuan
Wenli Ouyang
Changwen Zhang
Congrui Li
Yong Sun

Foundation Models (FMs) have demonstrated remarkable success in fields like computer vision and natural language processing, yet their application to combinatorial optimization remains underexplored. Optimization problems, often modeled as graphs, pose unique challenges due to their diverse structures, varying distributions, and NP-hard complexity. To address these challenges, we propose OPTFM, the first graph foundation model for general combinatorial optimization. OPTFM introduces a scalable multi-view graph transformer with hybrid self-attention and cross-attention to model large-scale heterogeneous graphs in $O(N)$ time complexity while maintaining semantic consistency throughout the attention computation. A Dual-level pre-training framework integrates node-level graph reconstruction and instance-level contrastive learning, enabling robust and adaptable representations at multiple levels. Experimental results across diverse optimization tasks show that models trained on OPTFM embeddings without fine-tuning consistently outperform task-specific approaches, establishing a new benchmark for solving combinatorial optimization problems.

PDF Details

AAAI Conference 2025 Conference Paper

ReactGPT: Understanding of Chemical Reactions via In-Context Tuning

Zhe Chen
Zhe Fang
Wenhao Tian
Zhaoguang Long
Changzhi Sun
Yuefeng Chen
Hao Yuan
Honglin Li

The interdisciplinary field of chemistry and artificial intelligence (AI) is an active area of research aimed at accelerating scientific discovery. Large language Models (LLMs) have shown significant promise in biochemical tasks, especially the molecule caption translation, which aims to align between molecules and natural language texts. However, existing works mainly focus on single molecules, while alignment between chemical reactions and natural language text remains largely unexplored. Additionally, the description of reactions is an essential part in biochemical patents and literature, and research on this aspect not only can help better understand chemical reactions but also promote research on automating chemical synthesis and retrosynthesis. In this work, we propose \textbf{ReactGPT}, a framework aiming to bridge the gap between chemical reaction and text. ReactGPT allows a new task: reaction captioning, by adapting LLMs to learn reaction-text alignment from context examples via In-Context Tuning. Specifically, ReactGPT jointly leverages a Fingerprints-based Reaction Retrieval module, a Domain-Specific Prompt Design module, and a two-stage In-Context Tuning module. We evaluate the effectiveness of ReactGPT on reaction captioning and experimental procedure prediction, both of these tasks can reflect the understanding of chemical reactions. Experimental results show that compared to previous models, ReactGPT exhibits competitive capabilities in resolving chemical reactions and generating high-quality text with correct structure.

PDF Details DOI

JMLR Journal 2021 Journal Article

DIG: A Turnkey Library for Diving into Graph Deep Learning Research

Meng Liu
Youzhi Luo
Limei Wang
Yaochen Xie
Hao Yuan
Shurui Gui
Haiyang Yu
Zhao Xu

Although there exist several libraries for deep learning on graphs, they are aiming at implementing basic operations for graph deep learning. In the research community, implementing and benchmarking various advanced tasks are still painful and time-consuming with existing libraries. To facilitate graph deep learning research, we introduce DIG: Dive into Graphs, a turnkey library that provides a unified testbed for higher level, research-oriented graph deep learning tasks. Currently, we consider graph generation, self-supervised learning on graphs, explainability of graph neural networks, and deep learning on 3D graphs. For each direction, we provide unified implementations of data interfaces, common algorithms, and evaluation metrics. Altogether, DIG is an extensible, open-source, and turnkey library for researchers to develop new methods and effortlessly compare with common baselines using widely used datasets and evaluation metrics. Source code is available at https://github.com/divelab/DIG. [abs] [ pdf ][ bib ] [ code ] &copy JMLR 2021. ( edit, beta )

PDF Details

AAAI Conference 2019 Conference Paper

Interpreting Deep Models for Text Analysis via Optimization and Regularization Methods

Hao Yuan
Yongjun Chen
Xia Hu
Shuiwang Ji

Interpreting deep neural networks is of great importance to understand and verify deep models for natural language processing (NLP) tasks. However, most existing approaches only focus on improving the performance of models but ignore their interpretability. In this work, we propose an approach to investigate the meaning of hidden neurons of the convolutional neural network (CNN) models. We first employ saliency map and optimization techniques to approximate the detected information of hidden neurons from input sentences. Then we develop regularization terms and explore words in vocabulary to interpret such detected information. Experimental results demonstrate that our approach can identify meaningful and reasonable interpretations for hidden spatial locations. Additionally, we show that our approach can describe the decision procedure of deep NLP models.

PDF Details

TCS Journal 2016 Journal Article

Average-case complexity of the min-sum matrix product problem

Ken C.K. Fong
Minming Li
Hongyu Liang
Linji Yang
Hao Yuan

We study the average-case complexity of min-sum product of matrices, which is a fundamental operation that has many applications in computer science. We focus on optimizing the number of “algebraic” operations (i. e. , operations involving real numbers) used in the computation, since such operations are usually expensive in various environments. We present an algorithm that can compute the min-sum product of two n × n real matrices using only O ( n 2 ) algebraic operations, given that the matrix elements are drawn independently and identically from some fixed probability distribution satisfying several constraints. This improves the previously best known upper-bound of O ( n 2 log ⁡ n ). The class of probability distributions under which our algorithm works include many important and commonly used distributions, such as uniform distributions, exponential distributions, folded normal distributions, etc. In order to evaluate the performance of the proposed algorithm, we performed experiments to compare the running time of the proposed algorithm with algorithms in [1]. The experimental results demonstrate that our algorithm achieves significant performance improvement over the previous algorithms.

Details DOI

SODA Conference 2010 Conference Paper

Data Structures for Range Minimum Queries in Multidimensional Arrays

Hao Yuan
Mikhail J. Atallah

Given a d -dimensional array A with N entries, the Range Minimum Query (RMQ) asks for the minimum element within a contiguous subarray of A. The 1D RMQ problem has been studied intensively because of its relevance to the Nearest Common Ancestor problem and its important use in stringology. If constant-time query answering is required, linear time and space preprocessing algorithms were known for the 1D case, but not for the higher dimensional cases. In this paper, we give the first linear-time preprocessing algorithm for arrays with fixed dimension, such that any range minimum query can be answered in constant time.

Details

TCS Journal 2007 Journal Article

Longest increasing subsequences in windows based on canonical antichain partition

Erdong Chen
Linji Yang
Hao Yuan

Given a sequence π 1 π 2 … π n, a longest increasing subsequence (LIS) in a window π 〈 l, r 〉 = π l π l + 1 … π r is a longest subsequence σ = π i 1 π i 2 … π i T such that l ≤ i 1 < i 2 < ⋯ < i T ≤ r and π i 1 < π i 2 < ⋯ < π i T. We consider the Lisw problem, which is to find the longest increasing subsequences in a sliding window of fixed-size w over a sequence. Formally, it is to find a LIS for every window in a set S FIX = { π 〈 i + 1, i + w 〉 ∣ 0 ≤ i ≤ n − w } ∪ { π 〈 1, i 〉, π 〈 n − i, n 〉 ∣ i < w }. By maintaining a canonical antichain partition in windows, we present an optimal output-sensitive algorithm to solve this problem in O ( output ) time, where output is the sum of the lengths of the n + w − 1 LISs in those windows of S FIX. In addition, we propose a more generalized problem called Lisset problem, which is to find a LIS for every window in a set S VAR containing variable-size windows. By applying our algorithm, we provide an efficient solution for the Lisset problem to output a LIS (or all the LISs) in every window which is better than the straightforward generalization of classical LIS algorithms. An upper bound of our algorithm on the Lisset problem is discussed.

Details DOI