Title | ||
---|---|---|
Deep Fault Prediction with Flexible Weighted Mining Based Alarm Correlation Analysis of Communication Networks |
Abstract | ||
---|---|---|
At present, most fault prediction methods of communication networks do not deeply mine the relationship between massive alarms, and are lack of deep feature engineering, so the prediction performance needs to be improved. We build a deep fault prediction model based on alarm correlation analysis and propose the feature interaction structure for feature engineering. For alarm correlation analysis, we propose flexible weighted sequential pattern mining (FWSPM) algorithm, which can flexibly determine the weight of alarms in different sequences and constrain the pruning process according to the weights. Then, we build a deep learning model based on feature interaction and introduce expert experience in the field of communication network management. Experimental results show that our model performs better in the historical alarms of the network management system than existing models. Besides, the proposed FWSPM method improves the performance of the proposed deep fault prediction model. |
Year | DOI | Venue |
---|---|---|
2020 | 10.1109/ICCT50939.2020.9295800 | 2020 IEEE 20th International Conference on Communication Technology (ICCT) |
Keywords | DocType | ISSN |
data mining,sequential pattern mining,deep learning | Conference | 2576-7844 |
ISBN | Citations | PageRank |
978-1-7281-8142-4 | 0 | 0.34 |
References | Authors | |
5 | 7 |
Name | Order | Citations | PageRank |
---|---|---|---|
Jun Jia | 1 | 0 | 0.34 |
Chunyan Feng | 2 | 305 | 38.57 |
Tiankui Zhang | 3 | 487 | 62.41 |
Hailun Xia | 4 | 45 | 12.02 |
Jinling Li | 5 | 0 | 0.34 |
Zhongtian Du | 6 | 0 | 0.34 |
Chenggang Li | 7 | 0 | 0.34 |