i am trying a time series classification task where i have to predict/classify in advance a driving scenario, making sure my predictions are accurate too and how ahead or time or delayed, it made a prediction. I am working with lstm model but it is too much overfitted. Accuracy curves for validation not changing, confusion matrix too is detecting only one label for each of my 3 labels. I tried SMOTE; sklearn class weights too, nothing improved, my ques is is there data augmentation for time series data too?.... and other problems too if some solutions or hints can be given.....
i am trying a time series classification task where i have to predict/classify in advance a driving scenario, making sure my predictions are accurate too and how ahead or time or delayed, it made a prediction. I am working with lstm model but it is too much overfitted. Accuracy curves for validation not changing, confusion matrix too is detecting only one label for each of my 3 labels. I tried SMOTE; sklearn class weights too, nothing improved, my ques is is there data augmentation for time series data too?.... and other problems too if some solutions or hints can be given.....
