5 Papers
1 Citations
Yaofeng Tu is an academic researcher from Nanjing University of Aeronautics and Astronautics. The author has contributed to research in topics: Computer science & Active learning (machine learning). The author has co-authored 5 publications. Previous affiliations of Yaofeng Tu include ZTE.
Chat about Author
Papers
Semi-supervised Time Series Anomaly Detection Model Based on LSTM Autoencoder
Hui Xiao,Donghai Guan,Rui Zhao,Weiwei Yuan,Yaofeng Tu,Asad Masood Khattak +5 more
- 20 Dec 2020
TL;DR: This work improves the loss function of the LSTM autoencoder so that it can be affected by unlabeled data and labeled data at the same time, and learn the distribution of unlabeling data andlabeled data atThe same time by minimizing the lossfunction.
4
Hard Disk Failure Prediction via Transfer Learning
Rui Zhao,Donghai Guan,Yuanfeng Jin,Hui Xiao,Weiwei Yuan,Yaofeng Tu,Asad Masood Khattak +6 more
- 20 Dec 2020
TL;DR: The DFPTL (Disk Failure Prediction via Transfer Learning) approach is proposed, which introduces the DANN approach to predict failure in heterogeneous disk systems by reducing the distribution differences between different models of disk datasets.
3
Posterior Transfer Learning with Active Sampling
Jie Pan,Yaofeng Tu +1 more
- 20 Dec 2020
TL;DR: A novel strategy to reuse the posterior probabilities from source domains without data sharing and a sampling strategy based on distribution difference is designed to actively select the most valuable instances for label querying is proposed.
1
Graph Representation Learning Using Attention Network
Bijay Gaudel,Donghai Guan,Weiwei Yuan,Deepanjal Shrestha,Chen Bing,Yaofeng Tu +5 more
- 22 Oct 2020
TL;DR: In this paper, a graph neighbor sampling, aggregation, and ATtention (GSAAT) framework is proposed to learn to aggregate the information of node's neighbors and stacked a layer in which nodes are able to attend over the aggregated information of their neighbors feature.
Cross-Task and Cross-Model Active Learning with Meta Features
Guo-Xiang Li,Yaofeng Tu,Sheng-Jun Huang +2 more
- 20 Dec 2020
TL;DR: In this article, a meta-feature based active learning method is proposed to predict the improvement of model performance for a candidate sample in a particular learning state, and the meta regressor is trained on the experience from previous active learning outcomes.