Search

• An end-to-end neural network-based framework for predicting seismic responses of building clusters. • The framework adopts the idea of embedding to fully characterize the earthquake and structure features and their interaction processes. • The framework applies to different prediction tasks such as pre-earthquake or post-earthquake assessments, and engineering scenarios with highly limited training data. • The framework achieves a balance of accuracy and efficiency with the representation of embeddings. Urban seismic damage assessment has recently become an emerging research topic due to the accelerating global urbanization trend, to which the seismic responses of building clusters to various earthquakes are a prerequisite. Traditional methods for this task, including vulnerability analysis and time-consuming time history analysis, may suffer from accuracy or efficiency problems especially for nonlinear response calculation. Machine learning methods allow for rapid and accurate response prediction, but current applications still lack scalability (on the size of structures or earthquakes) and the corresponding real datasets. To tackle this issue, this paper proposes an artificial neural network framework for simultaneously predicting nonlinear seismic responses of all buildings in a cluster subjected to multi-earthquake inputs. Inspired by the advanced collaborative filtering techniques, the framework converts the regional response prediction into a matrix completion problem, thereby aggregating information extracted from historical response records and physical characteristics to improve performance. The framework is used to assess nonlinear responses of a real urban region consisting of 2788 buildings subjected to 3798 measured ground motions. The results clearly demonstrated that the proposed framework achieves orders of magnitude faster than time history analysis and average errors below 3 % on several response metrics, showing high computational efficiency and accuracy. [ABSTRACT FROM AUTHOR]

Recommender Systems are a subclass of machine learning systems that employ sophisticated information filtering strategies to reduce the search time and suggest the most relevant items to any particular user. Hybrid recommender systems combine multiple recommendation strategies in different ways to benefit from their complementary advantages. Some hybrid recommender systems have combined collaborative filtering and content-based approaches to build systems that are more robust. In this paper, we propose a hybrid recommender system, which combines Alternating Least Squares (ALS) based collaborative filtering with deep learning to enhance recommendation performance as well as overcome the limitations associated with the collaborative filtering approach, especially concerning its " cold-start " problem. In essence, we use the outputs from ALS (collaborative filtering) to influence the recommendations from a Deep Neural Network (DNN), which combines characteristic, contextual, structural and sequential information, in a big data processing framework. We have conducted several experiments in testing the efficacy of the proposed hybrid architecture in recommending smartphones to prospective customers and compared its performance with other open-source recommenders. The results have shown that the proposed system has outperformed several existing hybrid recommender systems. [ABSTRACT FROM AUTHOR]

• Autoencoder can provide more accurate results in Collaborative Filtering. • Neural networks give satisfying results in rating prediction due to non-linearity. • Autoencoder has research potential in Collaborative Filtering and rating prediction. • The usage of Autoencoder and temporal signals jointly give remarkable RMSE scores. • With the advent of GPUs, neural networks become popular in Collaborative Filtering. Recommendation systems are used in various types of online platforms and in e-commerce. Collaborative filtering (CF) is one of the most popular approaches for recommendation systems and has been widely studied in academia. In recent years, several models based on neural networks that can discover nonlinear relationships have been proposed and compared to traditional CF models. The results showed that they performed better in terms of their prediction accuracy. However, these models do not consider user bias and item bias together, and they do not include temporal signals. This paper proposes a biased autoencoder model (Biased AutoRec) for CF, which is built on the well-known AutoRec CF approach. Several approaches are also proposed to integrate temporal signals into the Biased AutoRec model to merge the power of nonlinearity and temporal signals. Experiments on several public datasets showed that the new models outperformed the AutoRec model, which outperformed the prediction accuracy of previous state-of-the-art CF models (i.e., biased matrix factorization, RBM-CF, LLORMA). [ABSTRACT FROM AUTHOR]

• Deep learning-based aspect extraction model for building recommender system. • Exploiting wordembedding and POS tag embedding for aspect extraction. • Using LDA and lexicon-based method to generate aspect-based ratings. • Integration of the aspect ratings into tensor factorization technique for rating prediction. The main goal of Aspect-Based Opinion Mining is to extract product's aspects and the associated user opinions from the user text review. Although this serves as vital source information for enhancing rating prediction performance, few studies have attempted to fully utilize it for better accuracy of recommendation systems. Most of these studies typically assign equal weights to all aspects in the opinion mining process, however, in practices; users tend to give different priority on different aspects of the product when reaching overall ratings. In addition, most of the existing methods typically rely on handcrafted, rule-based or double propagation methods in the opinion mining process which are known to be time-consuming and often inclined to errors. This could affect the reliability and performance of the recommender systems (RS). Therefore, in this paper, we propose a weighted A spect-based O pinion mining using D eep learning method for R ecommender system (AODR) that can extract product's aspects and the underlying weighted user opinions from the review text using a deep learning method and then fuse them into extended collaborative filtering (CF) technique for improving the RS. The proposed method is basically comprised of two components: (1) Aspect-based opinion mining module which aims to extract the product aspects from the review text to generate aspect rating matrix. (2) Recommendation generation component that uses tensor factorization (TF) technique to compute weighted aspect ratings and finally infer the overall rating prediction. We evaluate the proposed model in terms of both aspect extraction and recommendation performance. Experiment results on different datasets show that our AODR model achieves better results compared to the baselines. [ABSTRACT FROM AUTHOR]