search

guillaume-chevalier / LSTM-Human-Activity-Recognition

Human Activity Recognition example using TensorFlow on smartphone sensors dataset and an LSTM RNN. Classifying the type of movement amongst six activity categories - Guillaume Chevalier
MIT License
3.33k stars 935 forks source link

How to format my own dataset for the RNN? #20

Open longtran84 opened 6 years ago

longtran84 commented 6 years ago

How to detect a cycle ?

guillaume-chevalier commented 6 years ago

42

longtran84 commented 6 years ago

@guillaume-chevalier

Iam having an issue, it is unable to detect the status when cycling, could you help me more detail how to do it? or make an example about structure data-set while cycling?

guillaume-chevalier commented 6 years ago

You'd basically need to refer to the original dataset's description and window your dataset like the one I used here, so that it fits with an amount of time steps, features, and batch_size examples during the training. You could change the dimensions as you wish to fit your data, and even use Truncated BPTT if you wish to keep states.

To understand the vocabulary better, refer to this: https://www.quora.com/What-do-samples-features-time-steps-mean-in-LSTM/answer/Guillaume-Chevalier-2?share=62ab4426&srid=C3n2