Is LSTM a Time Series Modeling Approach?

LSTM is also one way of doing time series modeling in a recurrent neural network, but it does it totally different. So what we do is we explicitly map a sequence of observations on an internal latent representation. And those internal representations we shape according to what we want. But our system uses unsupervised learning, whereas an LSTM or a recurrent neuralNetwork needs supervised learning. In our system, I only have a sequence of inputs. And from that, I try to explain future future states.