1
$\begingroup$

Trying to find the right search terms for literature on classifying 2D time series data.

I am looking at data from positional tracking of a swarm of insects over time. I have example datasets for specific insect behaviors, and I would like to play around with training a classifier on these.

$\endgroup$
0
$\begingroup$

You might be interested in RNNs (e.g., LSTMs) or Hidden Markov models. RNNs can be used directly for classification. HMMs can be used to build a generative model for each class, which can then be used for classification: if $x$ is the observed 2D time series, the model for class $c$ lets you compute $P(x|c)$, and then you can classify $x$ by selecting the class $c$ that maximizes $P(x|c)$ (the maximum likelihood principle).

$\endgroup$
  • $\begingroup$ HMMs are the approach I ended up going with. Thanks for the feedback! $\endgroup$ – Ryan Aug 16 '17 at 20:05
1
$\begingroup$

The one-nearest neighbor classifier is very competitive for time series. http://www.cs.ucr.edu/~eamonn/ICML2006.pdf If you want code or data, I have lots of both. eamonn

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.