Closed bbdamodaran closed 2 years ago
@bbdamodaran super late, but I only recently got around to put the code for the whole model online. Here it is: https://github.com/michetonu/DA-RNN_manoeuver_anticipation
My best advice is to try and optimize the parameter lambda – i.e. how much weight to give to each branch of the model. It is still not guaranteed to work for all data and problems though!
I implemented DANN in keras using your gradient reversal layer, but I don't good results. It will be nice, if could share your code
thanks