Cooperative Neural Network Training for Shared Vehicle Sensor Learning

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing vehicle communication systems lack sufficient bandwidth to effectively share and utilize information from other vehicles for timely decision-making, limiting the effectiveness of machine learning in predicting vehicle motion and preventing collisions.

Innovation Solution

Implementing cooperative learning neural networks that leverage advanced wireless communication protocols like 5G for high-bandwidth, low-latency data exchange between vehicles, allowing them to train and identify conditions or objects more accurately using combined sensor data from multiple sources.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If multiple independent neural networks are trained separately for different tasks, then each network can be optimized for its specific task, but the system requires multiple separate training processes and increases overall system complexity

Engineering Contradiction:
Improvetask-specific performanceVSAvoidsystem complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The patent combines multiple independent neural network training processes into a single cooperative learning system where multiple networks train simultaneously on different subsets of data. The networks share gradient information through a coordinator node, allowing them to learn different tasks concurrently within one unified training framework, thereby reducing system complexity while maintaining task-specific performance.

Inventive Principle:
Principle #5Merging (Combining)

Solution Approach 2:

The training data is segmented into different subsets assigned to different neural networks, allowing each network to focus on specific tasks or data portions. This segmentation enables parallel processing and distributed training while maintaining the ability to optimize for individual task requirements through the cooperative learning mechanism.

Inventive Principle:
Principle #1Segmentation

2Device complexity

If a single neural network is used for multiple tasks, then system complexity is reduced, but the network cannot be optimized for specific tasks and may suffer from task interference

Engineering Contradiction:
Improvesystem complexityVSAvoidtask-specific optimization
Core Design Contradiction:
Device complexityVSReliability

Solution Approach 1:

Each neural network in the cooperative learning system develops specialized local quality by training on specific data subsets and tasks. The networks maintain distinct weight configurations optimized for their respective tasks while still participating in the broader cooperative learning framework, allowing task-specific optimization without requiring a single monolithic network.

Inventive Principle:
Principle #3Local quality

3Measurement precision

If more training data is used to improve model accuracy, then learning precision increases, but the training time and computational resources increase proportionally

Engineering Contradiction:
Improvelearning precisionVSAvoidtraining time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The large training dataset is segmented into multiple smaller subsets that are distributed across different neural networks. Each network processes a portion of the data in parallel, reducing the training time for each individual network while collectively utilizing the entire dataset to maintain high learning precision across all tasks.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The cooperative learning system enables continuous useful action by having multiple networks train simultaneously on different data subsets rather than sequentially. This parallel processing approach maintains continuous learning progress across all networks, reducing overall training time while still processing the complete dataset.

Inventive Principle:
Principle #20Continuity of useful action

Data Source

PatentEP3677057B1Cooperative learning neural networks and systems
Publication Date: 2026.04.08 MICRON TECHNOLOGY INC
  • EP3677057B1 patent drawingFigure 1
  • EP3677057B1 patent drawingFigure 2
  • EP3677057B1 patent drawingFigure 3

AI summary

Systems, methods, and apparatuses related to cooperative learning neural networks are described. Cooperative learning neural networks may include neural networks which utilize sensor data received wirelessly from at least one other wireless communication device to train the neural network. For example, cooperative learning neural networks described herein may be used to develop weights which are associated with objects or conditions at one device and which may be transmitted to a second device, where they may be used to train the second device to react to such objects or conditions. The disclosed features may be used in various contexts, including machine-type communication, machine-to-machine communication, device-to-device communication, and the like. The disclosed techniques may be employed in a wireless (e.g., cellular) communication system, which may operate according to various standardized protocols.