Energy consumption management by novelty detection

Novelty detection approach is typically used when the quantity of available “abnormal” data is insufficient to construct explicit models for non-normal classes. Novelty detection has gained much research attention in application domains involving large datasets acquired from critical systems (Industry, Healthcare, Security, Video Surveillance, etc.).

Recent successful applications of Recurrent Neural Networks (RNNs) based on Long-Short Term Memory (LSTM)) have been demonstrated to be particularly useful for learning sequences containing longer term patterns of unknown length, due to their ability to maintain long term memory.

We analyzed the prediction performances of a Recurrent Neural Network architecture for energy consumption management varying the length of the processed input sequence and the size of the time window used in the feature extraction. Results corroborated the hypothesis that sequential models work better when dealing with data characterized by temporal order. However, so far the optimization of the temporal dimension remains an open issue. Finally we compared our approach with more traditional methods based on Gaussian distribution.

Tutti gli orari

Sunday 14 October
From 12.00 pm to 1.00 pm
Room 8 pav. 8
Energy consumption management by novelty detection

Antonio Rizzo, Giovanni Burresi

Antonio is a Full Professor of Cognitive Science & Technology at the Siena University. Psychologist and Tennist.

Category Talk & Conference · Type Talk