State-of-the-art models for joint entity recognition and relation extraction strongly rely on external natural language processing (NLP) tools such as POS (part-of-speech) taggers and dependency parsers. Thus, the performance of such joint models depends on the quality of the features obtained from these NLP tools. However, these features are not always accurate for various languages and contexts. In this paper, we propose a joint neural model which performs entity recognition and relation extraction simultaneously, without the need of any manually extracted features or the use of any external tool. Specifically, we model the entity recognition task using a CRF (Conditional Random Fields) layer and the relation extraction task as a multi-head selection problem (i.e., potentially identify multiple relations for each entity). We present an extensive experimental setup, to demonstrate the effectiveness of our method using datasets from various contexts (i.e., news, biomedical, real estate) and languages (i.e., English, Dutch). Our model outperforms the previous neural models that use automatically extracted features, while it performs within a reasonable margin of feature-based neural models, or even beats them.

No abstract

Abstract-The electrification of the vehicle fleet will result in an additional load on the power grid. Adequately dealing with such pluggable (hybrid) electrical vehicles (PHEV) forms part of the challenges and opportunities in the evolution towards Smart Grids. In this paper, we investigate the potential benefits of using control mechanisms, that could be offered by a Home Energy control box, in optimizing energy consumption stemming from PHEV charging in a residential use case. We present smart energy control strategies based on quadratic programming for charging PHEVs, aiming to minimize the peak load and flatten the overall load profile.We compare two strategies, and benchmark them against a business-as-usual scenario assuming full charging starting upon plugging in the PHEV. The first, local strategy only uses information at the home where the PHEV is charged: as a result the charging is optimized for local loads. The local strategy is compared to a global iterative strategy which controls the charging of multiple vehicles based on global load information over a residential area. Both strategies control the duration and rate of charging and result in charging schedules for each vehicle. We present quantitative simulation results over a set of 150 homes, and discuss the strategies in terms of complexity and performance (esp. resulting energy consumption), as well as their requirements concerning infrastructure and communication.

Non-intrusive load monitoring methods aim to disaggregate the total power consumption of a household into individual appliances by analysing changes in the voltage and current measured at the grid connection point of the household.The goal is to identify the active appliances, based on their unique fingerprint.An informative characteristic to attain this goal is the voltage-current trajectory.In this paper, a weighted pixelated image of the voltage-current trajectory is used as input data for a deep learning method: a convolutional neural network that will automatically extract key features for appliance classification. The macro-average F -measure is 77.60% for the PLAID dataset and 75.46% for the WHITED dataset.

No abstract

No abstract

No abstract

Adversarial training (AT) is a regularization method that can be used to improve the robustness of neural network methods by adding small perturbations in the training data. We show how to use AT for the tasks of entity recognition and relation extraction. In particular, we demonstrate that applying AT to a general purpose baseline model for jointly extracting entities and relations, allows improving the state-of-the-art effectiveness on several datasets in different contexts (i.e., news, biomedical, and real estate data) and for different languages (English and Dutch).