Auto-scaling Volatile Memory Allocation via Machine Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Managers of electronic networks and computing devices face challenges in determining and implementing efficient volatile memory allocations for metadata, as volatile memory is only retained as long as the device is powered on, leading to data loss when the device is turned off, necessitating a system for real-time, dynamic, and recorded memory allocations.
Innovation Solution
A system that uses a processing device with a volatile memory allocation machine learning model to access, analyze, and adjust metadata allocations in real-time, applying machine learning to determine optimal memory scaling and record changes for future use, even when the device is powered off, through a system that includes a modeler component for data format conversion and encryption for secure access.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If manual volatile memory allocation methods are used, then memory allocation can be performed, but the process is inefficient and difficult to manage for electronic networks and computing devices
Solution Approach 1:
The system employs machine learning models that automatically perform volatile memory allocation without human intervention. The ML model analyzes metadata, determines current allocations, and autonomously optimizes memory distribution based on learned patterns from historical data, enabling the system to serve itself rather than requiring manual management
Solution Approach 2:
The patent replaces manual mechanical memory allocation processes with an automated machine learning-based system. The ML model substitutes human managers and manual procedures with an intelligent algorithm that processes metadata, learns from historical allocation patterns, and automatically determines optimal memory allocations
2Speed
If volatile memory is used for metadata storage, then processing speeds are improved, but data is lost when the device is powered off
Solution Approach 1:
The system performs preliminary actions by training machine learning models on historical volatile memory allocation data while the device is operational. The trained model and its learned patterns are then saved to non-volatile storage, preserving the knowledge gained from historical data even when the device is powered off, allowing rapid resumption of optimized allocations
Solution Approach 2:
The patent creates copies of the learned allocation patterns and metadata from volatile memory to non-volatile storage. This copying mechanism ensures that valuable allocation knowledge is preserved across power cycles, allowing the system to recover and continue using optimized allocation strategies without losing the benefits of learned patterns
3Adaptability or versatility
If dynamic volatile memory allocation is implemented, then real-time optimization is achieved, but the system complexity increases
Solution Approach 1:
The system implements dynamic volatile memory allocation through machine learning models that continuously adapt to changing conditions. The ML model processes current metadata and historical patterns in real-time, dynamically adjusting memory allocations based on learned relationships between application behavior and optimal memory distribution
Solution Approach 2:
The patent changes the parameters of memory allocation by using machine learning to determine optimal allocation values based on metadata analysis. The system transforms static allocation parameters into dynamic, learned parameters that adapt to different applications and workloads, achieving real-time optimization through parameter transformation rather than complex manual configuration
4Measurement precision
If machine learning models are used for memory allocation, then accuracy and automation are improved, but computational resources are consumed
Solution Approach 1:
The system performs the computationally intensive machine learning model training in advance during periods when resources are available. The trained model is then saved and reused for making allocation decisions, avoiding the need to retrain the model repeatedly and reducing the computational burden during actual memory allocation operations
Solution Approach 2:
The patent implements periodic training of the machine learning model rather than continuous training. The model is trained at scheduled intervals or when significant new data becomes available, balancing the need for accurate predictions with the consumption of computational resources. Between training periods, the model makes predictions using its existing learned patterns
Data Source
AI summary
Systems, computer program products, and methods are described herein for auto-scaling volatile memory allocation in an electronic network. The present invention is configured to access metadata of at least one volatile memory component, wherein the metadata is associated with at least one application; determine a current volatile memory allocation for the metadata; determine a current metadata format of the metadata; apply the metadata to a volatile memory allocation machine learning model; generate, based on the application of the metadata to the volatile memory allocation machine learning model, a new volatile memory allocation for the metadata; and apply the new volatile memory allocation to the metadata of the at least one volatile memory component, wherein the application of the new volatile memory allocation to the metadata comprises at least one of an upscaling, a downscaling, or a constant.