How Does an LSTM Network Work?
LSTM is a recurrent neural network (RNN) architecture that REMEMBERS values over arbitrary intervals. LSTM is well-suited to classify, process and predict time series given time lags of unknown duration. Relative insensitivity to gap length gives an advantage to LSTM over alternative RNNs, hidden Markov models and other sequence learning methods.
![Image for post|861x226]
Recurrent neural networks work well on data that involves sequences of data that change over time. The short-term memory of an RNN can be combated using two variants of RNN. They are Long Short Term Memory (LSTM), and Grated RNN (GRNN). These two are capable of learning long-term dependencies using gates. These gates are different tensor operations that can learn what information to add or remove to the hidden state of the feedback loop.
The main difference between LSTM and GRNN is that GRNN has two gates (Update gate, Reset gate) to control its memory. The LSTM has three gates (Update gate, Reset gate, Forget gate.
The applications of RNN are natural language process, sentiment analysis, DNA sequence classification, speech recognition, and language translation.