Ting Lu

Prof. Ting Lu

Shanghai University of Finance and Economics

Research Focus & Metrics

Academic Career Length

6Years (2021-Present)

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

Representative Publications

View All Publications
18 May 2021 in Proceedings of the AAAI Conference on Artificial Intelligence
Verified
Journal Article
Open access
PDF file
BGBiyang Guo
Biyang Guo
SHSongqiao Han
Songqiao Han
XHXiao Han
Xiao Han
HHHailiang Huang
Hailiang Huang
TLTing Lu
Ting Lu
Representing the true label as one-hot vector is the common practice in training text classification models. However, the one-hot representation may not adequately reflect the relation between the instance and labels, as labels are often not completely independent and instances may relate to multiple labels in practice. The inadequate one-hot representations tend to train the model to be over-confident, which may result in arbitrary prediction and model overfitting, especially for confused datasets (datasets with very similar labels) or noisy datasets (datasets with labeling errors). While training models with label smoothing can ease this problem in some degree, it still fails to capture the realistic relation among labels. In this paper, we propose a novel Label Confusion Model (LCM) as an enhancement component to current popular text classification models. LCM can learn label confusion to capture semantic overlap among labels by calculating the similarity between instance and labels during training and generate a better label distribution to replace the original one-hot label vector, thus improving the final classification performance. Extensive experiments on five text classification benchmark datasets reveal the effectiveness of LCM for several widely used deep learning classification models. Further experiments also verify that LCM is especially helpful for confused or noisy datasets and superior to the label smoothing method.

12 June 2021 in Mobile Information Systems
Verified
Journal Article
Open access
PDF file
XZXuemei Zhao
Xuemei Zhao
TLTing Lu
Ting Lu
YDYonghui Dai
Yonghui Dai
With the development of big data technologies, usage-based insurance (UBI) has received considerable attention from insurance companies. UBI products focus on identifying the relationship between the individual driver’s risk and online channel behavior variables from Internet of Vehicles (IoV) data. Although omnichannel information integration has promoted the development of many industries, it has not been used to improve the accuracy of driver risk classification models in insurance industries. This paper investigates the role of combining different channel variables in improving the classification of driver’s risk. Specifically, several models, including logistic regression and three different data mining techniques (neural networks, random forests, and support vector machines), augmented with driving behavior data based on the IoV and offline consumer behavior data collected from 4S (Sale, Spare part, Service, Survey) dealers, are applied to the classification model of risk. The empirical results show that the inclusion of online and offline channel data improves the different risk assessments; results also demonstrate the importance of offline consumer behavior variables in different models. These insights have important implications for insurance companies on UBI pricing strategy and cost management.

1 July 2021 in Economics Letters
Verified
Journal Article
Closed access
PLPengfei Luo
Pengfei Luo
TLTing Lu
Ting Lu
DSDanDan Song
DanDan Song