ESP
ESP copied to clipboard
damellis
The Example-based Sensor Predictions (ESP) system applies machine learning to real-time sensor data.
I am under the impression that we take a snapshot of GRT and use it for most of the development. However, given the [PR](https://github.com/nickgillian/grt/pull/76) and other changes that has been...
It can be useful to have one function that checks individual calibration samples and another that does the overall computation of calibration settings once all the samples are collected. When...
I realize that training the model can take a long time, but maybe we can do it in the background? It seems like it might be simpler to have the...
It might be nice to contribute this to upstream so we can switch back to using ofTimeseriesPlot instead of the plotter class. See: https://github.com/nickgillian/ofxGrt/issues/8
Sometimes, it's useful to see the different dimensions of a feature vector on the same plot, in different colors, as with the live data stream. For instance, one dimension might...
That is, for x-axis (time) values where the feature isn't ready, leave that portion of the plot blank.
We should be able to pass in a matrix of highlight booleans and of labels when setting the entire data set for a time series plot.
That is, the shorter summary plot (with time as its x-axis) should be a spectrogram, rather than showing each dimension as a separate line. For these high-dimensional data sets, we...
This probably requires adding some sort of toString() method to the stream class, so that we can print a reasonable representation of the class that can't be opened.
So that we can score pipelines based on how it does relative to that ground truth information. This is probably complicated, as the form of ground truth data may change...