Events are the core element of information in descriptive corpus. Although many progresses have beenmade in Event Detection (ED), it is still a challenge in Natural Language Processing (NLP) to detect event information from data with unavoidable noisy labels. A robust Joint-training Graph ConvolutionNetworks (JT-GCN) model is proposed to meet the challenge of ED tasks with noisy labels in this paper. Specifically, we first employ two Graph Convolution Networks with Edge Enhancement (EE-GCN) tomake predictions simultaneously. A joint loss combining the detection loss and the contrast loss fromtwonetworks is then calculated for training. Meanwhile, a small-loss selection mechanism is introduced tomitigate the impact of mislabeled samples in networks training process. These two networks gradually reach an agreement on the ED tasks as joint-training progresses. Corrupted data with label noise are generated from the benchmark dataset ACE2005. Experiments on ED tasks has been conducted with bothsymmetry and asymmetry label noise on dif erent level. The experimental results show that the proposedmodel is robust to the impact of label noise and superior to the state-of-the-art models for EDtasks.

Artificial intelligence technology provides an unprecedented opportunity to assess the state of large-scale equipment with oil monitoring data. One of the key challenges in analyzing HFC (Hydraulic Fluid Composition) data is constructing a small sample classification, identifying abnormal equipment subgroups, and finding the significant impact indicators in unbalanced equipment. We propose GMBPN, a monitoring framework to identify the abnormal state and the order of influence index through multiple BP neural networks with group sampling. In order to improve the accuracy of small sample classification caused by the unbalanced number of samples, the classification model of small sample training is established by the quantitative grouping index. For the optimal classification model, the contribution order of each feature is compared by increased information gain. When GMBPN is applied to HFC data, it successfully captures the representative characteristics of abnormal equipment and impactors and shows its advantages over classical K-means and BP neural models in accuracy, classification consistency, and sampling methods.

No abstract