Multi-Platform Model Execution With Distributed Orchestration Monitoring
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing platforms lack the capability to provide a multi-platform modeling environment for executing models locally and on distributed systems efficiently.
Innovation Solution
A multi-platform model processing and execution management engine that can execute models internally or outsource execution to a distributed model execution orchestration engine, with a model data monitoring and analysis engine for monitoring and transmitting notifications.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If models are executed on distributed systems, then processing capability and scalability are improved, but system complexity increases
Solution Approach 1:
The patent introduces a model execution manager as an intermediary component that sits between the model development platform and the distributed execution environment. This manager handles the complexity of deploying models to distributed systems, managing compute resources, and coordinating execution without requiring users to directly interact with the complex distributed infrastructure. The manager abstracts away the complexity while enabling distributed processing capabilities.
2Adaptability or versatility
If multi-platform model execution is enabled, then versatility is improved, but device complexity increases
Solution Approach 1:
The patent implements a universal model execution manager that can handle multiple model formats and platforms through a single unified interface. The manager includes converters that automatically translate between different model formats (e.g., TensorFlow, PyTorch, ONNX) and platform-specific execution requirements. This multi-functional approach allows the system to execute models from various platforms without requiring separate specialized components for each platform, thereby reducing overall system complexity while maintaining versatility.
3Reliability
If real-time monitoring of model degradation is implemented, then reliability is improved, but use of energy increases
Solution Approach 1:
The patent implements periodic monitoring of model performance metrics rather than continuous real-time monitoring. The system samples model output at predetermined intervals to detect degradation patterns. This periodic approach reduces the computational overhead and energy consumption associated with constant monitoring, while still enabling timely detection of model degradation to trigger retraining or alerting when necessary.
Data Source
AI summary
Systems and methods are disclosed for managing the processing and execution of models that may have been developed on a variety of platforms. A multi-model execution module specifying a sequence of models to be executed may be determined. A multi-platform model processing and execution management engine may execute the multi-model execution module internally, or outsource the execution to a distributed model execution orchestration engine. A model data monitoring and analysis engine may monitor the internal and/or distributed execution of the multi-model execution module, and may further transmit notifications to various computing systems.