Sun Jian | Searcharxiv

arXiv2025

Research on Novelty Measurement Indicator of Academic Papers Based on the Atypical Recombination of Knowledge

The advancement of science is inherently dependent on the recombination of existing knowledge, and innovative research typically relies on the atypical recombination of established knoweldge bases. This study introduces a Knowledge Eccentricity to enable timely assessment of the novelty of research outputs by quantifying their degree of deviation from the existing knowledge system. For empirical analysis, we selected sample data including research articles published in Science and Nature, top 1% highly cited papers, and zero-cited papers for the year 2005, 2010, 2015, 2020, and 2025. We calculated the knowledge eccentricity scores for these papers and examined their potential influencing factors. The results indicate that team size exerts a significant negative effect on paper novelty, meaning larger team size is less conductive to enhancing the novelty of research outputs. Conversely, the number of references shows a signifcant positive correlation with paper novelty, which means that a greater number of references is associated with a moderate imporovement in a paper's novelty. The proposed indicator offers strong timeliness and operability, allowing for the evaluation of a paper's novelty immediately upon its publication.

cs.DL↗

arXiv2018

SFace: An Efficient Network for Face Detection in Large Scale Variations

Face detection serves as a fundamental research topic for many applications like face recognition. Impressive progress has been made especially with the recent development of convolutional neural networks. However, the issue of large scale variations, which widely exists in high resolution images/videos, has not been well addressed in the literature. In this paper, we present a novel algorithm called SFace, which efficiently integrates the anchor-based method and anchor-free method to address the scale issues. A new dataset called 4K-Face is also introduced to evaluate the performance of face detection with extreme large scale variations. The SFace architecture shows promising results on the new 4K-Face benchmarks. In addition, our method can run at 50 frames per second (fps) with an accuracy of 80% AP on the standard WIDER FACE dataset, which outperforms the state-of-art algorithms by almost one order of magnitude in speed while achieves comparative performance.

cs.CV↗