I’m trying to solve a problem where I’m I have IMU data for human activities that transition from one activity to the other based on time. What is the best approach for this? I tried svm but it doesn’t take into account the temporal properties of my features. What would be a good deep learning approach if any? Here I’m not trying to find the accuracy of classification but the point where one class was predicted from the other and how far was it from the actual transition. There are multiple classes based on activities which follow a pattern.