Modeling the Uncertainty in 2D Moving Target Selection
summary
A 2D Ternary‑Gaussian model predicts the mean and variance of endpoint distributions for circular moving targets from initial distance, target diameter and velocity, and is fit to mouse stylus and touch data with R2 ≈ 0.94. A Bayes classifier built from the predicted distributions reduces error versus boundary and comet baselines, yielding up to 4% speed, 41% accuracy gains when tested across mouse, stylus, finger input and realistic UIs.
AI-written summary · about
chapters (12)
- 0:00Introduction
- 0:27Target Selection
- 2:17Solution
- 3:55Modern Approach
- 5:17Defining the Task
- 7:32Modeling the Uncertainty
- 11:48Y Dimension Variance
- 12:46User Study
- 13:41Application
- 15:32Study
- 16:01Target Construction Accuracy
- 16:42Conclusion
related talks
19mPredicting Finger-Touch Accuracy Based on the Dual Gaussian Distribution Model
09mVariance and Distribution Models for Steering Tasks
19mInvestigating Effects of Post-Selection Feedback for Acquiring Ultra-Small Targets on Touchscreen
20mNext-Point Prediction Metrics for Perceived Spatial Errors
17mNext-Point Prediction for Direct Touch Using Finite-Time Derivative Estimation
21mFast Confidence Estimation: Regression models
25mDiscovering Your Model's Known Unknowns and Unknown Unknowns
28mCommon pitfalls leading to wrongly estimated model performance