First time here? Go to: Steps to Run
This is a tool to Visualize the Machine Learning concept of Support Vector Machines - Radial Basis Function. It trains a model with a dataset containing 2 distinct sets of data points using the SVM Radial kernel. The user can add more points to the data set and change training parameters to re-train the model.
It is a cross-platform web-based tool that can run on a web browser on both Laptops and Mobile Phones. It is an Image Marker-based AR application and uses the Kanji Image marker.
Hosted at: https://shah-deep2.github.io/SVM_AR
Inspired from: SVM Demo
Made with ❤️ at eCampus SJSU
- Go to the webpage: https://shah-deep2.github.io/SVM_AR
- Grant all the permissions asked including access to the camera. This is required.
- Point the camera towards the Kanji Image Marker such that it is fully visible in the camera.
- A pre-trained 3D model for SVM-RBF should appear on the screen on top of the Kanji Image Marker.
- On a mobile device, use your fingers to rotate and zoom in/out the model. On a laptop/computer, use the mouse cursor to rotate the model.
- Click the red/green buttons with a plus sign (+) to add new red/green data points to the model. Click again for every data point you want to add.
- Press the grey re-model button (below the red button) to re-train the ML model and re-render the plot.
- Click the (i) info button to hide/show the info related to the model.
- When model info is visible after pressing the (i) button, you can control the parameters C and σ with the help of a slider.
- Moving the sliders to change C or σ automatically re-trains the model.
- Play around with the tool by adding different data points and testing different parameters.
