Presentation 2019-06-21
Rethinking the Test Dataset Construction of Machine Learning for Mobile Application Identification
Takamitsu Iwai, Akihiro Nakao,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) Machine Learning based application identification has been intensively studied for applications such as application specific network slicing and zero-rating services. In machine learning, a test dataset, which is usually divided chronologically in the collected dataset, is used to evaluate a trained model. We advocate that it is often the case with mobile network analysis that this way of training and validation is irrelevant because overestimation of a trained model may occur when the data from one user is included both in training and in test dataset. In this paper, we propose to use IMEI to identify users and isolate test set from the dataset. We observe that conventional method overestimates by about 4% of accuracy on average and by 10% in the worst case compared to our evaluation using IMEI-based split method. In addition, our evaluation also shows the necessity of the IMEI instead of source IP for data isolation, as a single UE may be assign to multiple source IPs over time and thus source IP may not be a substitute for IMEI.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) Machine Learning / Application Identification
Paper # NS2019-40
Date of Issue 2019-06-13 (NS)

Conference Information
Committee OCS / PN / NS
Conference Date 2019/6/20(2days)
Place (in Japanese) (See Japanese page)
Place (in English) MALIOS(Morioka)
Topics (in Japanese) (See Japanese page)
Topics (in English) Photonic network system, Optical network operation and administration, Optical network design, Traffic engineering, Signaling, GMPLS, Inter-domain route control, Network monitoring, Optical amplifier and Optical repeater, Optical crossconect (OXC) and Optical add/drop multiplexer (OADM), Optical multiplexing/demultiplexing equipment, Optical signal processing, Optical terminal equipment, Digital signal processing and error correction, Optical communication measurement, Core/metro system, Seafloor transmission system, Optical access system and NG-PON, Ethernet, Optical transport network (OTN), Transmission monitoring and control, Optical transmission system design and tool, Mobile and Mobile-optical cooperation, etc.
Chair Joji Maeda(Tokyo Univ. of Science) / Takehiro Tsuritani(KDDI Research) / Yoshikatsu Okazaki(NTT)
Vice Chair / Haruki Ogoshi(Furukawa Electric) / Hideaki Furukawa(NICT) / Kohei Shiomoto(Tokyo City Univ.) / Akihiro Nakao(Univ. of Tokyo)
Secretary (NTT) / Haruki Ogoshi(Furukawa Electric) / Hideaki Furukawa(NTT) / Kohei Shiomoto(Univ. of Electr-Comm.) / Akihiro Nakao(Osaka Pref Univ.)
Assistant / Keijiro Suzuki(AIST) / Takahiro Kodama(Kagawa Univ) / Shinya Kawano(NTT)

Paper Information
Registration To Technical Committee on Optical Communication Systems / Technical Committee on Photonic Network / Technical Committee on Network Systems
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) Rethinking the Test Dataset Construction of Machine Learning for Mobile Application Identification
Sub Title (in English)
Keyword(1) Machine Learning
Keyword(2) Application Identification
1st Author's Name Takamitsu Iwai
1st Author's Affiliation University of Tokyo(UTokyo)
2nd Author's Name Akihiro Nakao
2nd Author's Affiliation University of Tokyo(UTokyo)
Date 2019-06-21
Paper # NS2019-40
Volume (vol) vol.119
Number (no) NS-92
Page pp.pp.29-34(NS),
#Pages 6
Date of Issue 2019-06-13 (NS)