Selectable Suggestions Using Provisional Models for Stable Updates
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing suggestion models in human-computer dialogs are often inadequate, leading to missed opportunities for efficient task completion and resource waste due to inadequate updating and potential inaccuracies, which can interrupt services and consume unnecessary computational resources.
Innovation Solution
Implement a system that selectively relies on additional machine learning models alongside a default suggestion model to generate selectable suggestions, allowing for testing and refinement without disrupting existing processes, by processing application content with both models and using user feedback to train the additional models.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If an updated suggestion model is deployed quickly to improve suggestion coverage, then the robustness and coverage of suggestions are improved, but accuracy decreases and service interruptions occur
Solution Approach 1:
The system performs preliminary actions by training the updated suggestion model in parallel with the default model before full deployment. This allows the model to be prepared in advance and tested, ensuring readiness while maintaining service stability through gradual rollout.
Solution Approach 2:
The system applies partial action by using the updated suggestion model only for generating selectable suggestions rather than replacing the default model entirely. This selective application allows improved coverage in specific areas while maintaining overall service stability through the default model's continued operation.
2Reliability
If the default suggestion model is used to maintain service stability, then service reliability is preserved, but suggestion accuracy and coverage are insufficient
Solution Approach 1:
The system merges the default suggestion model and the updated suggestion model into a unified architecture where both models coexist. The default model ensures service stability while the updated model provides enhanced accuracy and coverage for specific suggestion types, particularly for automated assistant actions.
Solution Approach 2:
The system implements multi-functionality by assigning different roles to different models: the default model handles general suggestion generation to ensure stability, while the updated model specifically targets automated assistant action suggestions to improve accuracy and coverage in that domain.
3Adaptability or versatility
If additional suggestion models are deployed to improve suggestion quality, then suggestion coverage is enhanced, but computational resource consumption increases
Solution Approach 1:
The system applies local quality by deploying the updated suggestion model selectively for specific tasks (generating automated assistant action suggestions) rather than using it for all suggestion generation. This targeted approach enhances coverage where needed while minimizing unnecessary computational resource consumption.
Solution Approach 2:
The system uses partial action by implementing lazy loading and on-demand invocation of the updated suggestion model. The model is only activated when specific conditions are met (e.g., when automated assistant actions are detected in the content), avoiding continuous resource consumption while maintaining improved coverage when needed.
Data Source
AI summary
Implementations set forth herein relate to selectively relying on additional suggestion model(s) when generating selectable suggestions, while also maintaining access to a default suggestion model. The selectable suggestions can be generated using one or more additional multi-domain machine learning (ML) models, which can be optionally available to the client application, regardless of whether a default suggestion model remains useful for generating suitable suggestions. In some implementations, as the client application employs various additional multi-domain ML models, a particular model can be identified as improving suggestions for the client application, at least based on user feedback and/or other data. The particular model can then be selected to replace and/or supplement the default suggestion model, in order to provide more accurate suggestions that, when selected, initialize actions that can preserve time and computational resources.