Xin Fu

Xin Fu

Auto-generated profile, based on publicly available information

Xin Fu

School of Management, Xiamen University, Xiamen 361005, China

Research Focus & Metrics

Academic Career Length

13Years (2014-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
28 June 2014 in Neural Computing and Applications
Verified
Journal Article
Closed access
PDF file
XFXin Fu
Xin Fu
YSYun Shen
Yun Shen

13 November 2024 in Information & Management
Verified
Journal Article
Closed access
QLQinfang Luo
Qinfang Luo
SCShun Cai
Shun Cai
NLNing Lv
Ning Lv
XFXin Fu
Xin Fu

14 November 2019 in IEEE Access
Verified
Journal Article
Open access
MCMengfan Chen
Mengfan Chen
KLKai Li
Kai Li
Defu ZhangDefu Zhang
Defu Zhang
LZLing Zheng
Ling Zheng
XFXin Fu
Xin Fu
Two dimensional strip packing problem implies a sort of NP-hard combinational optimization, where a set of rectangles are defined all for filling a strip with finite width and infinite height being minimized. Optimization taking no account of rotation and guillotine constraints remains problematic in solving the problem. This paper proposes a new variant of a hybrid heuristic algorithm for optimizing solutions of the strip packing problem. This algorithm involves three key improvements: the building of δ scoring rules for selecting rectangles, the use of the red-black trees that stores rectangle indices for quickly locating the most suitable rectangles, and the embedding of a hierarchical method into a random local search to implement an optimization solution. Comparative studies show that the proposed algorithm variant outperforms state-of-the-art algorithms on almost all benchmark datasets.


4 November 2026 in Symmetry
Verified
Journal Article
Open access
PDF file
LTLinjing Tang
Linjing Tang
ZYZhou Yiyang
Zhou Yiyang
XFXin Fu
Xin Fu
SSSong-Zhi Su
Song-Zhi Su
Music offers a stringent test of embodied intelligence because performance couples precise timing, morphology-dependent sound production, action-conditioned perception and interpersonal coordination. We conducted a systematic mapping survey with a piano-priority mechanism sample. Source screening retained 12,175 version-merged records, thematic screening retained 1944 and task-context screening advanced 831. Abstract and selected-method assessment advanced 241 technically relevant records, from which 96 research papers were selected for full-text coding. We propose three operational tests, a five-axis taxonomy and a separate L0–L4 ladder. The 96 independent full-text coder pairs yielded linear-weighted Cohen’s κ=0.939 for sensorimotor closure and κ=0.958 for embodiment level. Among 62 physical musical papers, 20 reported task-level acoustic feedback, 18 numerical system/component delay and 38 hardware limitation, failure or recovery; none of nine socially interactive physical papers separately quantified both influence directions. These are descriptive reporting frequencies in a purposive piano-priority sample and do not estimate field-wide prevalence. Forty-eight contextual references support definitions and transfer interpretation outside core denominators. The symmetry-aware synthesis separates geometric invariance/equivariance from directional partner influence and morphology-dependent action constraints. It supplies falsifiable propositions and proposed tests, rather than symmetry-coded prevalence results or new robot experiments.

15 November 2024 in IEEE Access
Verified
Journal Article
Open access
BXBingjie Xiang
Bingjie Xiang
JWJiarui Wu
Jiarui Wu
XHXiaoying Han
Xiaoying Han
QGQian Gu
Qian Gu
FCFei Chao
Fei Chao
XFXin Fu
Xin Fu
To address the issue of the substantial computational resource consumption during the inference phase of large language models due to their vast number of parameters, model sparsification is an effective solution. However, current sparsification methods for large models are costly. We propose a comprehensive two-stage approach called ELO-Mask for the rapid sparsification of large language models using a small calibration dataset. The approach consists of two steps: 1) Mask Reordering Step, this step involves initializing the mask using predefined parameter importance metrics, followed by reordering the model masks in blocks using the Straight-Through Estimator method with a small sample dataset. 2) Mask Fine-Tuning Step, this step involves further fine-tuning the masks obtained from the first step in blocks, using the same small sample dataset. Our experiments demonstrate the effectiveness of this approach. When sparsifying the Llama-7B model, our method shows significant superiority over the standard sparsification plus LoRA fine-tuning approach. It achieves comparable performance in the final sparse model while consuming less computational power, using a smaller dataset, occupying less GPU memory, and not affecting the inference speed of the sparse model.