Hey there! I am trying to substitute LSTM in time series prediction with Models because LSTM is time and memory heavy. I have already read quite about HMMs and how they work. Yet, in all of the materials Transition matrix, emission matrix and initial probabilities were known. In my case Hidden states are known (i want to use them as set) alongside with observations, but Transition and emission matrices and initial probabilities are unknown. Any ideas or hints how I can turn it into problem?

Source link
thanks you RSS link
( https://www.reddit.com/r//comments/91qkr6/d_hidden_markov__as_supervised_learning/)


Please enter your comment!
Please enter your name here