论文标题
机器学习与5G网络中的深度学习 - 科学影响的比较
Machine Learning vs. Deep Learning in 5G Networks -- A Comparison of Scientific Impact
论文作者
论文摘要
第五代（5G）无线网络技术的引入与新一代移动应用程序的高容量和速度需求的关键需求相匹配。人工智能（AI）的最新进展还赋予了5G蜂窝网络，其中两个主流是机器学习（ML）和深度学习（DL）技术。我们的研究旨在通过统计文献计量学手段来揭示这两种技术对这两种技术的差异。执行的分析包括有关索引类型，资金可用性，期刊或会议发布选项的引文表现，以及这些指标的分布，以详细的方式评估流行趋势。 Web of Science（WOS）数据库主机2245纸，用于ML和1407篇DL相关研究的论文。 DL研究以2013年的9％开始，在所有与DL和ML相关的研究中，在2022年达到45％。与科学影响相关的结果表明，与5G的ML研究（2.118）相比，DL研究的平均标准化引文略高于平均，而双方的SCI扩展索引论文往往具有相似的引用性能（分别为3.165和3.162）。与DL相比，在ESCI中索引的ML相关研究显示了两次引用性能。在DL领域的会议论文和ML领域的期刊论文在科学利益方面比其差异较小的同行具有优越性。在2014年实现了ML研究的最高引用性能，而2017年DL研究则观察到了这一峰值。我们可以得出结论，与DL相关论文的出版率和引文率往往会通过引用指标在5G域中提高和胜过基于ML的研究。
Introduction of fifth generation (5G) wireless network technology has matched the crucial need for high capacity and speed needs of the new generation mobile applications. Recent advances in Artificial Intelligence (AI) also empowered 5G cellular networks with two mainstreams as machine learning (ML) and deep learning (DL) techniques. Our study aims to uncover the differences in scientific impact for these two techniques by the means of statistical bibliometrics. The performed analysis includes citation performance with respect to indexing types, funding availability, journal or conference publishing options together with distributions of these metrics along years to evaluate the popularity trends in a detailed manner. Web of Science (WoS) database host 2245 papers for ML and 1407 papers for DL-related studies. DL studies, starting with 9% rate in 2013, has reached to 45% rate in 2022 among all DL and ML-related studies. Results related to scientific impact indicate that DL studies get slightly more average normalized citation (2.256) compared to ML studies (2.118) in 5G, while SCI-Expanded indexed papers in both sides tend to have similar citation performance (3.165 and 3.162 respectively). ML-related studies those are indexed in ESCI show twice citation performance compared to DL. Conference papers in DL domain and journal papers in ML domain are superior in scientific interest to their counterparts with minor differences. Highest citation performance for ML studies is achieved for year 2014, while this peak is observed for 2017 for DL studies. We can conclude that both publication and citation rate for DL-related papers tend to increase and outperform ML-based studies in 5G domain by the means of citation metrics.
