Asynchronous Coagent Recommendations for Edge Synchronization Delays
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional recommendation systems in edge computing environments suffer from inefficiencies, inflexibilities, and inaccuracies due to synchronous algorithms that require geographic synchronization, leading to slow communication and resource consumption, and often fail to accurately represent client device preferences.
Innovation Solution
An edge-based recommender system utilizing an asynchronous coagent network with local coagents that learn and act cooperatively across edges, generating recommendations using local policy gradients and delayed scores, refining them as they are received, and updating parameters independently.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If synchronous algorithms are used for recommendation generation across edges, then coordination between edges is achieved, but communication time and resource consumption increase significantly
Solution Approach 1:
The patent segments the recommendation generation process into independent edge-level operations. Each edge computes recommendations autonomously using local data and models without waiting for other edges, eliminating the need for synchronous coordination while maintaining recommendation quality through distributed independent processing
Solution Approach 2:
The patent applies preliminary action by pre-training coagent models and preparing recommendation algorithms at each edge before runtime. This allows edges to immediately generate recommendations using pre-configured models and local data without requiring real-time communication or synchronization with other edges during the recommendation generation process
2Reliability
If synchronous algorithms are used for recommendation generation, then edge coordination is maintained, but computing resource consumption increases
Solution Approach 1:
Each edge serves itself by maintaining local coagent models and computing recommendations independently using local data and resources. This self-service approach eliminates the need for edges to consume computing resources waiting for or communicating with other edges, significantly reducing overall system resource consumption while maintaining coordination through distributed autonomous operation
Solution Approach 2:
The patent implements local quality by enabling each edge to have its own specialized coagent models trained on local data characteristics. This allows each edge to optimize recommendation generation locally without requiring centralized coordination or additional computing resources from other edges, reducing overall system resource consumption while maintaining recommendation accuracy
3Stability of the object's composition
If rigid synchronization between edges is enforced, then algorithm consistency is maintained, but system flexibility decreases when synchronization fails
Solution Approach 1:
The patent applies dynamics by making the system adaptable to different operational conditions. When edges are available, the system uses distributed coagent models for enhanced recommendations; when edges are unavailable or synchronization fails, the system gracefully degrades to using only local edge models. This dynamic behavior maintains algorithm consistency through standardized interfaces while providing flexibility to handle various failure scenarios
Solution Approach 2:
The patent implements beforehand cushioning by designing the system to handle synchronization failures in advance. Each edge is equipped with local coagent models that can independently generate recommendations without external coordination. This preparatory measure ensures that when synchronization fails due to network issues or edge unavailability, the system continues to function with reduced but still effective recommendations, maintaining flexibility without compromising algorithm consistency
4Adaptability or versatility
If bag-of-algorithms approach is used for recommendations, then various learning algorithms are leveraged, but theoretical grounding is lacking leading to inaccurate recommendations
Solution Approach 1:
The patent applies composite materials by creating a hybrid recommendation system that combines multiple learning algorithms (collaborative filtering, content-based filtering, deep learning) into a unified coagent framework with strong theoretical grounding. This composite approach integrates the strengths of different algorithms while providing a coherent mathematical foundation through coagent theory, achieving both algorithm diversity and recommendation accuracy that neither approach could achieve alone
Data Source
AI summary
The present disclosure relates to systems, methods, and non-transitory computer-readable media that generate digital item recommendations for client devices utilizing coagent recommendation models of a distributed asynchronous coagent network. Indeed, in one or more embodiments, the disclosed systems operate on an edge computing device of a distributed asynchronous coagent network. In some cases, the disclosed systems utilize recommendation scores generated at the edge computing device via local coagents and additional recommendation scores received from other coagents of other edge computing devices to generate a digital item recommendation. In some cases, the disclosed systems progressively refines the recommendation as delayed scores from the other coagents are received. Further, in some embodiments, the disclosed systems update parameters of the local coagents using local policy gradients determined from responses to the generated recommendations.