Bo Fu
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Bo Fu has a total of 55 co-authors. Here are the top co-authors by number of shared publications:
- Pei Liu
School of Computer Science and Engineering University of Electronic Science and Technology of China Xiyuan Ave 611731 Chengdu Sichuan China
9 shared publications - Feng Ye
Sichuan University
5 shared publications - Hong Zheng
Breast Center and Multi-omics Laboratory of Breast Diseases, West China Hospital, Sichuan University, Chengdu, China
4 shared publications - Luping Ji
School of Computer Science and Engineering, University of Electronic Science and Technology of China, Chengdu, China
4 shared publications - Hong Bu
Institute of Clinical Pathology, Key Laboratory of Transplantation Engineering and Immunology, Ministry of Health, West China Hospital, Sichuan University
3 shared publications - Junlin Zhu
Laboratory of Intelligent Collaborative Computing, University of Electronic Science and Technology of China, Chengdu, China
3 shared publications
Latest Publications
- Open Access
- 19 July 2026
- Published by Association for Computing Machinery (ACM)
- p. 2578-2588
Towards Conflict-aware Selective Knowledge Unlearning for Continual Few-shot Knowledge Graph Completion
- 19 July 2026
- Published by Association for Computing Machinery (ACM)
- p. 2578-2588
Open AccessTowards Conflict-aware Selective Knowledge Unlearning for Continual Few-shot Knowledge Graph Completion
- Open Access
- 19 June 2026
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Intelligent Systems and Technology
- Vol. 17 (5) , 1-25
CurvGCL: Mixed-Curvature Manifolds Graph Contrastive Learning for Knowledge-enhanced Recommendation
Recently, contrastive learning (CL) has gained significant traction in the recommendation field for its ability to extract stable signals from unsupervised data. However, practical applications often face challenges due to sparse and noisy data, limiting the effectiveness of CL in downstream tasks. Although knowledge graphs (KGs) offer valuable auxiliary information in such scenarios, they inherently possess significant non-Euclidean properties. Existing methods that embed data in a single constant curvature manifold can only accommodate specific graph structures, such as linear, hierarchical, or cyclic ones, making it challenging to model diverse and complex geometric topologies effectively. To address these limitations, we design CurvGCL, a novel graph contrastive learning framework that leverages mixed-curvature manifold representations to model complex geometric structures for knowledge-enhanced recommendation. Specifically, we propose a multi-manifolds adaptive fusion mechanism that preserves multiple geometric structures within KGs, enhancing their representation capabilities and allowing the integrated knowledge to adapt flexibly to various non-Euclidean geometric forms. Additionally, we design a novel method for adaptive-enhanced graph structure representation learning based on mixed-curvature manifold spaces, dynamically removing edges that are not relevant to recommendation during data augmentation, thereby generating high-quality views. Extensive experiments on four public datasets demonstrate that CurvGCL outperforms state-of-the-art methods.
- 19 June 2026
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Intelligent Systems and Technology
- Vol. 17 (5) , 1-25
Open AccessCurvGCL: Mixed-Curvature Manifolds Graph Contrastive Learning for Knowledge-enhanced Recommendation
Recently, contrastive learning (CL) has gained significant traction in the recommendation field for its ability to extract stable signals from unsupervised data. However, practical applications often face challenges due to sparse and noisy data, limiting the effectiveness of CL in downstream tasks. Although knowledge graphs (KGs) offer valuable auxiliary information in such scenarios, they inherently possess significant non-Euclidean properties. Existing methods that embed data in a single constant curvature manifold can only accommodate specific graph structures, such as linear, hierarchical, or cyclic ones, making it challenging to model diverse and complex geometric topologies effectively. To address these limitations, we design CurvGCL, a novel graph contrastive learning framework that leverages mixed-curvature manifold representations to model complex geometric structures for knowledge-enhanced recommendation. Specifically, we propose a multi-manifolds adaptive fusion mechanism that preserves multiple geometric structures within KGs, enhancing their representation capabilities and allowing the integrated knowledge to adapt flexibly to various non-Euclidean geometric forms. Additionally, we design a novel method for adaptive-enhanced graph structure representation learning based on mixed-curvature manifold spaces, dynamically removing edges that are not relevant to recommendation during data augmentation, thereby generating high-quality views. Extensive experiments on four public datasets demonstrate that CurvGCL outperforms state-of-the-art methods.
- Open Access
- 17 June 2026
- Published by Elsevier in SSRN Electronic Journal
Set-Valued Prediction for Large Language Models with Feasibility-Aware Coverage Guarantees
Large language models (LLMs) inherently operate over a large generation space, yet conventional usage typically reports the most likely generation (MLG) as a point prediction, which underestimates the model's capability: although the top-ranked response can be incorrect, valid answers may still exist within the broader output space and can potentially be discovered through repeated sampling. This observation motivates moving from point prediction to set-valued prediction, where the model produces a set of candidate responses rather than a single MLG. In this paper, we propose a principled framework for set-valued prediction, which provides feasibility-aware coverage guarantees. We show that, given the finite-sampling nature of LLM generation, coverage is not always achievable: even with multiple samplings, LLMs may fail to yield an acceptable response for certain questions within the sampled candidate set. To address this, we establish a minimum achievable risk level (MRL), below which statistical coverage guarantees cannot be satisfied. Building on this insight, we then develop a data-driven calibration procedure that constructs prediction sets from sampled responses by estimating a rigorous threshold, ensuring that the resulting set contains a correct answer with a desired probability whenever the target risk level is feasible. Extensive experiments on six language generation tasks with five LLMs demonstrate both the statistical validity and the predictive efficiency of our framework.
- 17 June 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessSet-Valued Prediction for Large Language Models with Feasibility-Aware Coverage Guarantees
Large language models (LLMs) inherently operate over a large generation space, yet conventional usage typically reports the most likely generation (MLG) as a point prediction, which underestimates the model's capability: although the top-ranked response can be incorrect, valid answers may still exist within the broader output space and can potentially be discovered through repeated sampling. This observation motivates moving from point prediction to set-valued prediction, where the model produces a set of candidate responses rather than a single MLG. In this paper, we propose a principled framework for set-valued prediction, which provides feasibility-aware coverage guarantees. We show that, given the finite-sampling nature of LLM generation, coverage is not always achievable: even with multiple samplings, LLMs may fail to yield an acceptable response for certain questions within the sampled candidate set. To address this, we establish a minimum achievable risk level (MRL), below which statistical coverage guarantees cannot be satisfied. Building on this insight, we then develop a data-driven calibration procedure that constructs prediction sets from sampled responses by estimating a rigorous threshold, ensuring that the resulting set contains a correct answer with a desired probability whenever the target risk level is feasible. Extensive experiments on six language generation tasks with five LLMs demonstrate both the statistical validity and the predictive efficiency of our framework.
- Open Access
- 10 November 2025
- Published by Association for Computing Machinery (ACM)
- p. 4519-4528
DebiasedKGE: Towards Mitigating Spurious Forgetting in Continual Knowledge Graph Embedding
- 10 November 2025
- Published by Association for Computing Machinery (ACM)
- p. 4519-4528
Open AccessDebiasedKGE: Towards Mitigating Spurious Forgetting in Continual Knowledge Graph Embedding
- 2 November 2023
- Published by Elsevier in Medical Image Analysis
- Vol. 91, 103020
AdvMIL: Adversarial multiple instance learning for the survival analysis on whole-slide images
- 2 November 2023
- Published by Elsevier in Medical Image Analysis
- Vol. 91, 103020
AdvMIL: Adversarial multiple instance learning for the survival analysis on whole-slide images
- 28 April 2023
- Published by Elsevier in Expert Systems with Applications
- Vol. 227
DSCA: A dual-stream network with cross-attention on whole-slide image pyramids for cancer prognosis
- 28 April 2023
- Published by Elsevier in Expert Systems with Applications
- Vol. 227
DSCA: A dual-stream network with cross-attention on whole-slide image pyramids for cancer prognosis
- Open Access
DSCA: A Dual-Stream Network with Cross-Attention on Whole-Slide Image Pyramids for Cancer Prognosis
The cancer prognosis on gigapixel Whole-Slide Images (WSIs) has always been a challenging task. To further enhance WSI visual representations, existing methods have explored image pyramids, instead of single-resolution images, in WSIs. In spite of this, they still face two major problems: high computational cost and the unnoticed semantical gap in multi-resolution feature fusion. To tackle these problems, this paper proposes to efficiently exploit WSI pyramids from a new perspective, the dual-stream network with cross-attention (DSCA). Our key idea is to utilize two sub-streams to process the WSI patches with two resolutions, where a square pooling is devised in a high-resolution stream to significantly reduce computational costs, and a cross-attention-based method is proposed to properly handle the fusion of dual-stream features. We validate our DSCA on three publicly-available datasets with a total number of 3,101 WSIs from 1,911 patients. Our experiments and ablation studies verify that (i) the proposed DSCA could outperform existing state-of-the-art methods in cancer prognosis, by an average C-Index improvement of around 4.6%; (ii) our DSCA network is more efficient in computation -- it has more learnable parameters (6.31M vs. 860.18K) but less computational costs (2.51G vs. 4.94G), compared to a typical existing multi-resolution network. (iii) the key components of DSCA, dual-stream and cross-attention, indeed contribute to our model's performance, gaining an average C-Index rise of around 2.0% while maintaining a relatively-small computational load. Our DSCA could serve as an alternative and effective tool for WSI-based cancer prognosis.
Open AccessDSCA: A Dual-Stream Network with Cross-Attention on Whole-Slide Image Pyramids for Cancer Prognosis
The cancer prognosis on gigapixel Whole-Slide Images (WSIs) has always been a challenging task. To further enhance WSI visual representations, existing methods have explored image pyramids, instead of single-resolution images, in WSIs. In spite of this, they still face two major problems: high computational cost and the unnoticed semantical gap in multi-resolution feature fusion. To tackle these problems, this paper proposes to efficiently exploit WSI pyramids from a new perspective, the dual-stream network with cross-attention (DSCA). Our key idea is to utilize two sub-streams to process the WSI patches with two resolutions, where a square pooling is devised in a high-resolution stream to significantly reduce computational costs, and a cross-attention-based method is proposed to properly handle the fusion of dual-stream features. We validate our DSCA on three publicly-available datasets with a total number of 3,101 WSIs from 1,911 patients. Our experiments and ablation studies verify that (i) the proposed DSCA could outperform existing state-of-the-art methods in cancer prognosis, by an average C-Index improvement of around 4.6%; (ii) our DSCA network is more efficient in computation -- it has more learnable parameters (6.31M vs. 860.18K) but less computational costs (2.51G vs. 4.94G), compared to a typical existing multi-resolution network. (iii) the key components of DSCA, dual-stream and cross-attention, indeed contribute to our model's performance, gaining an average C-Index rise of around 2.0% while maintaining a relatively-small computational load. Our DSCA could serve as an alternative and effective tool for WSI-based cancer prognosis.
- 20 February 2023
- Published by Elsevier in Computer Methods and Programs in Biomedicine
- Vol. 231, 107433
GraphLSurv: A scalable survival prediction network with adaptive and sparse structure learning for histopathological whole-slide images
- 20 February 2023
- Published by Elsevier in Computer Methods and Programs in Biomedicine
- Vol. 231, 107433
GraphLSurv: A scalable survival prediction network with adaptive and sparse structure learning for histopathological whole-slide images
- 3 May 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Biomedical and Health Informatics
- Vol. 26 (8) , 4258-4269
Dynamic Sepsis Prediction for Intensive Care Unit Patients Using XGBoost-Based Model With Novel Time-Dependent Features
Sepsis is a systemic inflammatory response caused by pathogens such as bacteria. Because its pathogenesis is not clear, the clinical manifestations of patients vary greatly, and the alarming incidence and mortality pose a great threat to patients and medical systems, especially in the ICU (Intensive Care Unit). The traditional judgment criteria have the problem of low specificity. Artificial intelligence models could greatly improve the accuracy of sepsis prediction and judgment. Based on the XGBoost machine learning framework taking demographic, vital signs, laboratory tests and medical intervention data as input, this paper proposes a novel model for dynamically predicting sepsis and assessing risk. To realize the model, two methods for feature construction are introduced. For the observed time-series data of vital signs and laboratory tests, the time-dependent method performs to construct the time-dependent characteristics after the statistical screening. For the clinical intervention data, the statistical counting method is applied to construct count-dependent characteristics. Moreover, a new objective function is proposed for the XGBoost framework, and the first-order and second-order gradients of the objective function are also given for model training. Compared with the state-of-the-art methods at present, the proposed model has the best performance, with AUROC improved by 5.4% on the MIMIC-III dataset and 2.1% on PhysioNet Challenge 2019 dataset. The data processing and training methods of this model can be conveniently applied in different electronic health record systems and has a wide application prospect.
- 3 May 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Biomedical and Health Informatics
- Vol. 26 (8) , 4258-4269
Dynamic Sepsis Prediction for Intensive Care Unit Patients Using XGBoost-Based Model With Novel Time-Dependent Features
Sepsis is a systemic inflammatory response caused by pathogens such as bacteria. Because its pathogenesis is not clear, the clinical manifestations of patients vary greatly, and the alarming incidence and mortality pose a great threat to patients and medical systems, especially in the ICU (Intensive Care Unit). The traditional judgment criteria have the problem of low specificity. Artificial intelligence models could greatly improve the accuracy of sepsis prediction and judgment. Based on the XGBoost machine learning framework taking demographic, vital signs, laboratory tests and medical intervention data as input, this paper proposes a novel model for dynamically predicting sepsis and assessing risk. To realize the model, two methods for feature construction are introduced. For the observed time-series data of vital signs and laboratory tests, the time-dependent method performs to construct the time-dependent characteristics after the statistical screening. For the clinical intervention data, the statistical counting method is applied to construct count-dependent characteristics. Moreover, a new objective function is proposed for the XGBoost framework, and the first-order and second-order gradients of the objective function are also given for model training. Compared with the state-of-the-art methods at present, the proposed model has the best performance, with AUROC improved by 5.4% on the MIMIC-III dataset and 2.1% on PhysioNet Challenge 2019 dataset. The data processing and training methods of this model can be conveniently applied in different electronic health record systems and has a wide application prospect.
- Open Access
- 18 February 2022
- Published by Association for Computing Machinery (ACM)
- p. 67-73
DeepGCNMIL: Multi-head Attention Guided Multi-Instance Learning Approach for Whole-Slide Images Survival Analysis Using Graph Convolutional Networks
- 18 February 2022
- Published by Association for Computing Machinery (ACM)
- p. 67-73
Open AccessDeepGCNMIL: Multi-head Attention Guided Multi-Instance Learning Approach for Whole-Slide Images Survival Analysis Using Graph Convolutional Networks