ESP
ESP copied to clipboard
damellis
The Example-based Sensor Predictions (ESP) system applies machine learning to real-time sensor data.
This may need to be done directly in ofApp, since FirmataStream can't inherit from BaseSerialInputStream.
i.e. a way to get a call to a function like `prediction(label)` or `result(float[])` instead of having to do the serial parsing manually.
I think we can check Classifier::getSupportsNullRejection().
Instead of, say, replacing spaces with underscores while the user is typing the name.
During training, the GRT DTW implementation picks one template from each class and only compares live data to those templates. It would be useful to show the user which sample...
For both confusion / overlap and impact / information gain. Needed for #267 and #264.
Our current approach based on class likelihoods may not be the best for DTW. Maybe borrow the goodness score from MAGIC? This should probably be for both the confusion /...
Currently, we have a generic classifier that's based on the information gain of the sample (i.e. the negative log of the probability of classifying the sample correctly). This can, I...
http://jafari.tamu.edu/wp-content/uploads/2015/12/IEEESensorsPaper2014_10.pdf
On the calibration tab, we may want to show (or enable) a button that takes the user to the training tab after they've finished collecting any needed calibration samples.