An Improved Elman Neural Network for Daily Living Activities Recognition

Zaineb Liouane (1) , Tayeb Lemlouma (2) , Philippe Roose (3) , Frédéric Weis (4) , Messaoud Hassani (1)

1 ENIM - École Nationale d’Ingénieurs de Monastir
2 IRISA - Institut de Recherche en Informatique et Systèmes Aléatoires
3 LIUPPA - Laboratoire Informatique de l'Université de Pau et des Pays de l'Adour
4 TACOMA - TAngible COMputing Architectures

One of the main issues regarding the monitoring of persons in a smart home environment is the accuracy of the daily control of the person, the health prevention and the timely prediction of abnormal situations. To tackle this problem, this work proposes the use of an improved version of the Elman Neural Network (Elman-NN). In order to minimize the error between inputs and desired outputs, we optimize some criteria of the network to gain good results. We propose to use the Differential Evolution algorithm in the learning step of the Elman-NN to evolve the error performance. Our proposed model is responsible to predict the activities of the monitored elderly and to detect any state changes. This hybridization will help to optimize the weight and the bias of the network to achieve our objective function and to obtain a good network. The experimental results reveal that the proposed model is satisfactory for elderly person’s movement prediction.

Mots clés: Elderly Care; Elman Neural Network; Differential Evolution algorithm; Prediction; Smart spaces

Domaines: Réseaux et télécommunications [cs.NI]; Intelligence artificielle [cs.AI]; Apprentissage [cs.LG]; Réseau de neurones [cs.NE]; Santé publique et épidémiologie

Zaineb Liouane, Tayeb Lemlouma, Philippe Roose, Frédéric Weis, Messaoud Hassani. An Improved Elman Neural Network for Daily Living Activities Recognition. 16th International Conference on Intelligent Systems Design and Applications, Dec 2016, Porto, Portugal. pp.697-707, ⟨10.1007/978-3-319-53480-0_69⟩. ⟨hal-01462982⟩