Search

Scientific quantitative description of air quality uncertainty remains a challenging task. The key to accurate and interpretable forecasting of air quality lies in combining spatial and temporal information and feature interpretability. This study proposes a novel spatio-temporal forecasting model based on deep learning with squeeze and excitation attention mechanism and interpretable feature selection strategy. Specifically, the data is collected from Beijing-Tianjin-Hebei region which is highly representative and convenient for accuracy comparison. And interpretable feature selection model effectively combines the muti-stage feature selection and Shapley additive explanation methods to identify individual features that contribute significantly to the forecasting target as base variable. Subsequently, a deterministic forecasting model based on improved deep learning with squeeze and excitation channel attention to adaptively adjust feature weights is constructed to implement air quality forecasting. Finally, feature interpretable model analyzes the changes in different air quality influencing factors as time advances, providing policy guidance for joint prevention and control of regional air pollution. Experimental and discussion results reveals that the proposed model can serve as an effective air pollution management and analysis tool.

Abstract Value at risk (VaR) and expected shortfall (ES) have emerged as standard measures for detecting the market risk of financial assets and play essential roles in investment decisions, external regulations, and risk capital allocation. However, existing VaR estimation approaches fail to accurately reflect downside risks, and the ES estimation technique is quite limited owing to its challenging implementation. This causes financial institutions to overestimate or underestimate investment risk and finally leads to the inefficient allocation of financial resources. The main purpose of this study is to use machine learning to improve the accuracy of VaR estimation and provide an effective tool for ES estimation. Specifically, this study proposes a VaR estimator by combining quantile regression with “Mogrifier” recurrent neural networks to capture the “long memory” and “clustering” properties of financial assets; while for estimating ES, this study directly models the quantile of assets and employs generative adversarial networks to generate future tail risk scenarios. In addition to the typical properties of financial assets, the model design is also consistent with heterogeneous market theory. An empirical application to four major global stock indices shows that our model is superior to other existing models.

Accurate prediction of shale gas production is important for the government to formulate energy policies. China’s shale gas production is characterized by limited data and unclear influencing factors. Grey prediction models hold advantages in solving problems with inadequate data and insufficient information. However, existing grey models for shale gas production prediction have drawbacks of insufficient characteristic expression and parameter optimization. Therefore, a new grey model with four parameters is proposed in this work. Firstly, the power index, accumulative order, and background value coefficient in the new model are optimized using the PSO algorithm. Then, based on an in-depth analysis of the model structure and tests using random samples, the new model’s compatibility and adaptability are demonstrated to be better than other similar grey models. Thirdly, the new model is applied to simulate China’s shale gas production. The results show that the comprehensive error of the new model is only 1.3626%, lower than that of the SGGM(1,1) and UGM(1,1) models, which are 1.6211% and 1.6575% respectively. Lastly, using the new model, China's shale gas production is predicted, which is expected to reach 752 hundred million cubic meters by 2030. After analyzing the rationality of the result, relevant suggestions are put forward.

This paper addresses the problem of cyber-attack detection in a discrete event system by proposing a novel model. The model utilizes graph convolutional networks to extract spatial features from event sequences. Subsequently, it employs gated recurrent units to re-extract spatio-temporal features from these spatial features. The obtained spatio-temporal features are then fed into an attention model. This approach enables the model to learn the importance of different event sequences, ensuring that it is sufficiently general for identifying cyber-attacks, obviating the need to specify attack types. Compared with traditional methods that rely on synchronous product computations to synthesize diagnosers, our deep learning-based model circumvents state explosion problems. Our method facilitates real-time and efficient cyber-attack detection, eliminating the necessity to specifically identify system states or distinguish attack types, thereby significantly simplifying the diagnostic process. Additionally, we set an adjustable probability threshold to determine whether an event sequence has been compromised, allowing for customization to meet diverse requirements. Experimental results demonstrate that the proposed method performs well in cyber-attack detection, achieving over 99.9% accuracy at a 1% threshold and a weighted F1-score of 0.8126, validating its superior performance.