The examples below are implemented in CodeCircle. Click the purple “>>” icon to hide the source code.
This example demonstrates classification and regression models from RapidLib. The models are trained using predefined training data at the top of the script, and then incoming mouse positions are evaluated. The evaluation simulates an XOR-like function — 0 is in the upper-left corner, 1 is in the upper-right and lower-left corners, and 2 is in the lower-right corner.
This example lets you train a classification model by associating mouse positions with numbers and colours.
This example lets you train a regression model by associating mouse positions with numbers and colours.
RapidLib_004: Series Classification
Dynamic Time warping can classify a series of events, and compare series of different lengths. In this example, two example shapes can be drawn and then a test shape is classified. Order is taken into account, so a clockwise circle would not match a counter-clockwise circle.
Currently only working in Chrome.
Mano_001: Mouse Shapes Recognition
Behind the scenes, mano sends the recorded examples to the como server, which provides an http based training service on a single API endpoint.
A trained model (a hierarchical hidden markov model in this case) is sent back in response, allowing mano to perform client-side recognition.
To play with the example, type a key to set the current label (e.g “Z”), then draw some examples of the corresponding character on the topmost canvas. Type another key, then repeat again to train the model with more and more examples of different characters. Try the recognition on the second canvas below.
To reset the training set, reload the page.
Repovizz2 storing and retrieving training sets
Basic example that demonstrates how to store and retrieve training datasets using the repovizz2 RESTful API.
Waves-lfo_001: Moving Average Filter