Improving energy efficiency management has become an important task for current electricity market participating entities,and monitoring consumption of pivotal appliances plays an important role in many applications.T...Improving energy efficiency management has become an important task for current electricity market participating entities,and monitoring consumption of pivotal appliances plays an important role in many applications.This paper focuses on detecting whether a residence possesses a certain type of appliance based on their electricity consumption and the problem of class imbalance within deep learning model training for large power appliances with the state‘ON’.We propose a datadriven deep learning approach with attention mechanism to detect residential appliances from low-resolution aggregate energy consumption data.Firstly,the historical consumption profile of each user is divided into a specific length and labeled with the status of an appliance to generate training and test samples.Then,a deep convolutional neural network model with attention mechanism is trained,and the trained model is utilized to classify the test samples.Meanwhile,we obtain appliance status in a residence based on classification of multiple samples.Finally,we propose a novel approach of data generation for class imbalance of appliance detection using generative adversarial networks.In order to guarantee the quality,we devise a mechanism of self-validation to ensure generated data approximating real distribution of minor class samples.Experiments are conducted on a low-frequency smart meter data set sampled once every 30 minutes,and the results show that the proposed model performs better than hidden Markov model based algorithms and has good application prospects.展开更多
Scene recognition is a fundamental task in computer vision,which generally includes three vital stages,namely feature extraction,feature transformation and classification.Early research mainly focuses on feature extra...Scene recognition is a fundamental task in computer vision,which generally includes three vital stages,namely feature extraction,feature transformation and classification.Early research mainly focuses on feature extraction,but with the rise of Convolutional Neural Networks(CNNs),more and more feature transformation methods are proposed based on CNN features.In this work,a novel feature transformation algorithm called Graph Encoded Local Discriminative Region Representation(GEDRR)is proposed to find discriminative local representations for scene images and explore the relationship between the discriminative regions.In addition,we propose a method using the multi-head attention module to enhance and fuse convolutional feature maps.Combining the two methods and the global representation,a scene recognition framework called Global and Graph Encoded Local Discriminative Region Representation(G2ELDR2)is proposed.The experimental results on three scene datasets demonstrate the effectiveness of our model,which outperforms many state-of-the-arts.展开更多
基金This project is supported by“the Fundamental Research Funds for the Central Universities”N2017001。
文摘Improving energy efficiency management has become an important task for current electricity market participating entities,and monitoring consumption of pivotal appliances plays an important role in many applications.This paper focuses on detecting whether a residence possesses a certain type of appliance based on their electricity consumption and the problem of class imbalance within deep learning model training for large power appliances with the state‘ON’.We propose a datadriven deep learning approach with attention mechanism to detect residential appliances from low-resolution aggregate energy consumption data.Firstly,the historical consumption profile of each user is divided into a specific length and labeled with the status of an appliance to generate training and test samples.Then,a deep convolutional neural network model with attention mechanism is trained,and the trained model is utilized to classify the test samples.Meanwhile,we obtain appliance status in a residence based on classification of multiple samples.Finally,we propose a novel approach of data generation for class imbalance of appliance detection using generative adversarial networks.In order to guarantee the quality,we devise a mechanism of self-validation to ensure generated data approximating real distribution of minor class samples.Experiments are conducted on a low-frequency smart meter data set sampled once every 30 minutes,and the results show that the proposed model performs better than hidden Markov model based algorithms and has good application prospects.
基金This research is partially supported by the Programme for Professor of Special Appointment(Eastern Scholar)at Shanghai Institutions of Higher Learning,and also partially supported by JSPS KAKENHI Grant No.15K00159.
文摘Scene recognition is a fundamental task in computer vision,which generally includes three vital stages,namely feature extraction,feature transformation and classification.Early research mainly focuses on feature extraction,but with the rise of Convolutional Neural Networks(CNNs),more and more feature transformation methods are proposed based on CNN features.In this work,a novel feature transformation algorithm called Graph Encoded Local Discriminative Region Representation(GEDRR)is proposed to find discriminative local representations for scene images and explore the relationship between the discriminative regions.In addition,we propose a method using the multi-head attention module to enhance and fuse convolutional feature maps.Combining the two methods and the global representation,a scene recognition framework called Global and Graph Encoded Local Discriminative Region Representation(G2ELDR2)is proposed.The experimental results on three scene datasets demonstrate the effectiveness of our model,which outperforms many state-of-the-arts.
