18363277. Attribution and Generation of Saliency Visualizations for Machine-Learning Models simplified abstract (GOOGLE LLC)
Attribution and Generation of Saliency Visualizations for Machine-Learning Models
Organization Name
Inventor(s)
Andrei Kapishnikov of Watertown MA (US)
[[:Category:Fernanda Bertini Vi�gas of Lexington MA (US)|Fernanda Bertini Vi�gas of Lexington MA (US)]][[Category:Fernanda Bertini Vi�gas of Lexington MA (US)]]
Michael Andrew Terry of Cambridge MA (US)
Tolga Bolukbasi of Cambridge MA (US)
Attribution and Generation of Saliency Visualizations for Machine-Learning Models - A simplified explanation of the abstract
This abstract first appeared for US patent application 18363277 titled 'Attribution and Generation of Saliency Visualizations for Machine-Learning Models
Simplified Explanation
The patent application describes methods, systems, devices, and computer-readable media for saliency visualization.
- Data input with multiple features is received and segmented into regions.
- Attribution scores are generated for each feature, indicating their saliency.
- Gain values for regions are determined based on attribution scores over iterations.
- Regions with the greatest gain values are added to a saliency mask at each iteration.
- Saliency visualization is produced based on the saliency mask.
Potential Applications
- Image and video processing - Object detection and recognition - Medical imaging analysis
Problems Solved
- Identifying important features in complex data sets - Enhancing decision-making processes - Improving visual representation of data
Benefits
- Enhanced understanding of data - Improved accuracy in feature selection - Streamlined data analysis and visualization
Original Abstract Submitted
Methods, systems, devices, and tangible non-transitory computer readable media for saliency visualization are provided. The disclosed technology can include receiving a data input including a plurality of features. The data input can be segmented into regions. At least one of the regions can include two or more of the features. Attribution scores can be respectively generated for features of the data input. The attribution scores for each feature can be indicative of a respective saliency of such feature. A respective gain value for each region can be determined over one or more iterations based on the respective attribution scores associated with the features included in the region. Further, at each iteration one or more of the regions with the greatest gain values can be added to a saliency mask. Furthermore, at each iteration a saliency visualization can be produced based on the saliency mask.
