One of the newest and most challenging areas of artificial intelligence is generating new data by machines with learning the patterns in existing data. Among the types of data sequential ones are more complex, due to the dependencies that exist between their subsequence, and thus we require different methods for processing them. The most wellknown methods used to study and analyze these data are recurrent neural networks, But simple recurrent neural networks are not generative models, and for using them in task of generation, they need to be implemented in the form of a generative model. In this study, we introduce and implemente recurrent neural networks in the form of a developed model of variational autoencoder. This model is able to generate meaningful and new, intervals of time steps, in addition this is a multifunctional model which can be used simultaneously for three important tasks in the field of machine learning: 1) conditionally generating data: This means that we can generate new sequential data by considering different Key Words : Variational Autoencoder, Sequential Data, Neural Networks, Deep Learning, Variational Methods, Unsupervised Learning, EEG Signal, Rehabilitation
