Demand side management (DSM) programs are an integral part of the modern grid. Most of these DSM programs are designed to work at household and hour level. The optimization problems in these DSM programs are guided by the forecasted load. An error in the hour ahead load forecasting results in a suboptimal solution entailing economic cost to both the utility and the customers. Predicting loads at a fine granularity (e.g., households) is challenging due to a large number of (known or unknown) factors affecting power consumption. At larger scales (e.g., clusters of consumers), since the inherent stochasticity and fluctuations are averaged out, the problem becomes substantially easier. Many techniques have been proposed to predict loads of clusters of consumers in various localities with great accuracy. Also, these techniques generally utilize sociological and weather information and work better on data from a particular locality. In this paper, a new technique called Past Vector Similarity (PVS) has been proposed to predict electricity load one hour ahead at the level of an individual household. The proposed strategy is based on load information only and does not require calendar, weather or any other attributes. In fact, the idea is to extract the exact load patterns of individual households corresponding to their routine and socioeconomic values. Consequently, the technique makes use of the recent past vector and generate similar patterns for the prediction of future load profiles. Furthermore, the ensemble of these similar loads is an efficient prediction of electricity. PVS has just two parameters due to which it can be applied to the smaller dataset without overfitting issue. Moreover, due to the parallel nature of PVS, it can be scaled for a large number of customers without computation burden. The proposed PVS has been assessed empirically for two distinct datasets from Australia and Sweden. The simulation results demonstrate that the PVS significantly outperforms other state-of-the-art forecasting methods in terms of accuracy.

No abstract

Recent Deep Neural Networks (DNNs) based edge detection methods claim to beat human performance on small scale datasets like BSDS500. But is the problem of edge detection really solved? To answer this question, we review the existing dataset limitations as well as the generalization capabilities of the proposed architectures. To this end, we develop a Synthetic Textured Masks Dataset (STMD) that contains 28,000 gray scale images. The performance of several edge detection methods is severely degraded on STMD. To further validate these results we propose a baseline Single Scale Feed Forward Edge Detector (SFED), which is a simple 9-layer feed-forward convolutional neural network with no pooling layers. The performance of SFED is better than most state-of-the-art architectures on BSDS500 and is superior to all the compared architectures on STMD. These results show that most of the architectural advancements of existing architectures are at the cost of generalizability where if we change the dataset set distribution (both training and testset), the performance become significantly degraded and therefore the problem of edge detection is still far away from being solved. There are also severe limitations of existing datasets in the field, and STMD provides a framework for designing and testing better edge detection architectures for novel application areas, such as, medical imaging and self-driving cars.