Abstract Electricity price has been the focus of electricity market participants from the beginning of electricity markets as the price reflects the prevailing market conditions. Its importance and volatile behavior have motivated many research works in the area of electricity price forecasting. While qualified electricity price forecasting methods have been presented in the literature, prediction of price spikes, as a distinctive and crucial aspect of electricity price time series, has been less researched and still remained as a critical challenge. In this paper, a new deep learning-based framework is proposed for price spike occurrence and value prediction. Three deep generative models, for data generation, rebalanced data clustering, and point value forecasting, are combined in this framework. In the first step, the Rebalancing Variational Autoencoder (RebalVAE) is proposed to rebalance the data using a mixture of discrete and continuous variables. After that, the Clustering Variational Auto-Encoder (ClusVAE) with a new clustering-specific loss term from the mutual information theory is proposed to make an accurate and automatic clustering. Subsequently, a deep diffusion model is applied to each cluster for price value prediction. This model, namely Diffused Predictor Variational Auto Encoder (DiffPredVAE), includes a Transformer Encoder, embedded in both Encoder and Decoder networks of a VAE, and is equipped with a new loss function. Results on two real-world datasets (in five cases) confirm the superiority of the proposed price spike prediction model compared to various other models.

Social media plays a significant role in sharing essential information, which helps humanitarian organizations in rescue operations during and after disaster incidents. However, developing an efficient method that can provide rapid analysis of social media images in the early hours of disasters is still largely an open problem, mainly due to the lack of suitable datasets and the sheer complexity of this task. In addition, supervised methods can not generalize well to novel disaster incidents. In this paper, inspired by the success of Knowledge Distillation (KD) methods, we propose an unsupervised deep neural network to detect and localize damages in social media images. Our proposed KD architecture is a feature-based distillation approach that comprises a pre-trained teacher and a smaller student network, with both networks having similar GAN architecture containing a generator and a discriminator. The student network is trained to emulate the teacher’s behavior on training input samples, which, in turn, contain images that do not include any damaged regions. Therefore, the student network only learns the distribution of no damage data and would have different behavior from the teacher network facing damages. To detect damage, we utilize the difference between features generated by two networks using a defined score function that demonstrates the probability of damages occurring. Our experimental results on the benchmark dataset confirm that our approach outperforms state-of-the-art methods in detecting and localizing the damaged areas, especially for novel disaster types1.