Bilateral Gesture Training Model for Touch Input Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing machine-learned models for inferring gestures from user input often fail to accurately recognize intended gestures, leading to degraded user experience due to frequent negative inferences.
Innovation Solution
The system includes a touch sensor and one or more computing devices configured to input sensor data to a machine-learned model. It generates training data by annotating sensor data from failed attempts as positive examples of a gesture once a successful inference is made, and uses this data to retrain the model.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a machine-learned model is used to infer gestures from sensor data, then gesture recognition capability is provided, but recognition accuracy deteriorates leading to frequent negative inferences
Solution Approach 1:
The system performs preliminary actions by collecting and storing sensor data from failed gesture attempts before the model can learn from them. This preparatory data collection enables subsequent training iterations to incorporate previously unusable failure cases, improving recognition accuracy without compromising reliability.
Solution Approach 2:
The system implements feedback by using successfully recognized gestures to generate training data that annotates previously failed attempts as positive examples. This feedback loop continuously improves the model's accuracy by incorporating real user interaction patterns, transforming unreliable failure data into valuable training signals.
2Measurement precision
If sensor data from failed attempts is annotated and used for training, then model accuracy improves, but data processing complexity increases
Solution Approach 1:
The system applies self-service by automatically annotating failed gesture attempts using subsequent successful gesture recognitions. The model trains itself on its own failure data without requiring external intervention or complex manual labeling processes, simplifying the overall data processing workflow while improving accuracy.
Solution Approach 2:
The system inverts the traditional approach by treating failed attempts not as errors to be discarded but as valuable training data. By annotating failures as positive examples based on subsequent successful recognitions, the system transforms what was previously wasted data into the core mechanism for improving accuracy.
3Measurement precision
If the model is retrained with personalized training data, then user-specific recognition accuracy improves, but training time and computational resources increase
Solution Approach 1:
The system applies partial action by incrementally updating the model with small batches of newly collected training data from user interactions, rather than performing complete retraining. This approach achieves personalized accuracy improvement over time while minimizing training time and computational resource consumption at any given moment.
Data Source
AI summary
The present disclosure provides computer-implemented methods, systems, and devices for efficient bilateral training of users and devices with touch input systems. An interactive object generates, based on a first output of the machine-learned model in response to sensor data associated with a first touch input, first inference data indicating a negative inference corresponding to a first gesture. The interactive object generates, based on an output of the machine-learned model in response to sensor data associated with a second touch input, second inference data indicating a positive inference corresponding to the first gesture. The interactive object, in response to generating the positive inference subsequent to the negative inference, generates training data as a positive training example of the first gesture. The interactive object trains the machine-learned model based at least in part on the training data.