Federated Hyperdimensional Computing Model Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Federated Hyperdimensional Computing (HDC) models require significant resources for training, leading to resource constraints that compromise predictive performance, especially in scenarios with limited computational, wireless, energy, and storage resources.
Innovation Solution
A Resource-Efficient Federated Hyperdimensional Computing (RE-FHDC) framework divides the full-sized HDC model into multiple smaller sub-models, allowing for independent training on edge devices with reduced computational and communication costs, and aggregates these sub-models to form the full-sized model, enabling iterative training and inference.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a full-sized HDC model is trained in federated learning, then predictive performance is improved, but resource consumption (computational, energy, communication) increases significantly
Solution Approach 1:
The full-sized HDC model is divided into multiple smaller sub-models, each trained independently on edge devices. This segmentation reduces the computational burden and energy consumption on resource-constrained devices while maintaining the ability to achieve high predictive performance through aggregation of sub-model results.
2Measurement precision
If a full-sized HDC model is trained in federated learning, then predictive performance is improved, but communication cost increases
Solution Approach 1:
By segmenting the model into sub-models trained locally on edge devices, the communication overhead is reduced. Only sub-model parameters or aggregated results need to be transmitted rather than the entire full-sized model, thereby lowering communication costs while preserving predictive performance.
3Use of energy by moving object
If model size is reduced to meet resource constraints, then resource consumption is lowered, but predictive performance is sacrificed
Solution Approach 1:
The approach segments the full-sized model into multiple smaller sub-models that can be trained independently on resource-constrained edge devices. This enables the system to operate within resource constraints while maintaining predictive performance through the collective contribution of multiple sub-models.
4Productivity
If multiple smaller sub-models are trained instead of one full-sized model, then resource efficiency is improved, but model complexity increases
Solution Approach 1:
The full model training task is segmented into multiple independent sub-model training tasks that can be executed in parallel on different edge devices. This segmentation improves resource efficiency by distributing the computational load while the independence of sub-model training actually simplifies the overall process coordination.
Data Source
AI summary
A device to train a hyperdimensional computing (HDC) model may include memory and processing circuitry to train one or more independent sub models of the HDC model and transmit the one or more independent sub models to another computing device, such as a server. The device may be one of a plurality of devices, such as edge computing devices, edge or Internet of Things (IoT) nodes, or the like. Training of the one or more independent sub models of the HDC model may include transforming one or more training data points to one or more hyperdimensional representations, initializing a prototype using the hyperdimensional representations of the one or more training data points, and iteratively training the initialized prototype.