On: LSTM for Human Activity Recognition

[nharada]

Activity trackers like Fitbit can do this to some extent

[zump → nharada]

No gyro, not enough information.

[GChevalier → zump]

I also tried my code without the gyro, and incredibly, the accuracy is still pretty high (86.90% rather than 90.77%).

[zump → GChevalier]

Have you tried your classifier with a different accelerometer dataset? That will be a good test of generalization.

[GChevalier → zump]

I am co-writing a paper with a Ph.D. student and he is currently working on trying with other datasets. We are also trying different architectures, combining multiple LSTMs (stacked, residual connections + batch normalization, bidirectional LSTMs, and on)