Multi-task Network Model Communication via Fusion Feature Compression
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Implementing device-cloud collaboration in multi-task learning (MTL) models is challenging due to the complex structure and high computational requirements of multi-task network models, which necessitate efficient feature transmission and processing across devices.
Innovation Solution
A multi-task network model-based communication method that involves processing input signals using a backbone network model to obtain fusion features, performing compression and channel coding on these features, and transmitting them between communication apparatuses, facilitating device-cloud collaboration by reducing transmission overheads and enhancing anti-noise performance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If a multi-task network model is used to improve task resolution efficiency and reduce storage costs, then the model structure becomes more complex and requires more computing resources
Solution Approach 1:
The multi-task network model is divided into a backbone network model that extracts shared features and multiple functional network models that perform specific tasks. This segmentation allows the complex model to be broken down into manageable components, reducing the perceived complexity while maintaining the efficiency benefits of multi-task learning.
Solution Approach 2:
The backbone network model serves as a universal feature extractor that provides shared intermediate features to multiple functional network models. This multi-functional design allows a single model structure to handle multiple tasks simultaneously, improving productivity while managing complexity through resource sharing.
2Loss of substance
If a multi-task network model is deployed on mobile devices to reduce storage costs, then computing resources on mobile devices become insufficient
Solution Approach 1:
A communication apparatus is introduced as an intermediary between the mobile device and the cloud. The communication apparatus receives intermediate features from the mobile device, performs additional processing, and returns results. This intermediary approach allows the system to leverage cloud computing resources while maintaining the storage efficiency benefits of multi-task learning on mobile devices.
3Productivity
If the multi-task network model is divided between mobile device and cloud for device-cloud collaboration, then transmission overheads increase due to complex model structures
Solution Approach 1:
Only the essential intermediate features extracted by the backbone network model are transmitted between the mobile device and the cloud, rather than transmitting the entire complex model structure or raw data. This extraction approach minimizes transmission overhead while maintaining the collaborative processing benefits between device and cloud.
Data Source
AI summary
The technology of this application relates to a multi-task network model-based communication method, apparatus, and system. The method may include a first communication apparatus processing an input signal by using a first backbone network model, to obtain a fusion feature, where the fusion feature is obtained by fusing a plurality of first features which are obtained by performing feature extraction on the input signal. The method may further include performing compression and channel coding on the fusion feature to obtain first information, and sending the first information to a second communication apparatus. The second communication apparatus receives second information from the first communication apparatus, performs decompression and channel decoding on the second information to obtain a reconstructed feature of the fusion feature, performs feature parsing on the reconstructed feature by using a second backbone network model, to obtain a feature parsing result, and processes the feature parsing result by using a functional network model.