Multi-task gating for machine learning systems
By incorporating gating mechanisms in convolutional neural networks, the system dynamically learns to share or segregate features optimally, addressing the inefficiencies and interference in existing multi-task machine learning systems and enhancing both efficiency and accuracy.
WO2025064055A1 Publication Date: 2025-03-27QUALCOMM INC
0 Cites 0 Cited by
Patent Information
- Application Number
- PCT/US2024/039651
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- Priority Date
- 2023-09-18
- Filing Date
- 2024-07-25
- Publication Date
- 2025-03-27
AI Technical Summary
Technical Problem
Existing multi-task machine learning systems face challenges in efficiently training and inferring multiple tasks simultaneously, leading to redundant computation and suboptimal performance due to interference between tasks.
Method used
The implementation of gating mechanisms in convolutional neural networks allows for dynamic learning of optimal feature sharing and segregation schemes, enabling the neural network to determine whether to process shared or task-specific features based on a learnable gating mechanism.
Benefits of technology
This approach enhances the computational efficiency and accuracy of multi-task learning by allowing the network to adaptively select the most appropriate features for each task, reducing interference and improving overall performance.
✦ Generated by Eureka AI based on patent content.
Abstract
Systems and techniques are described herein for training and using multitask machine learning models. For example, a computing device can obtain training data for a first task in a layer in a neural network; perform, based on a determination from a first gating mechanism, the shared function on shared features of the training data using at least one shared channel to generate a shared feature map; perform, based on the determination from the first gating mechanism, the first task-specific function on first task-specific features of the training data using at least one first task-specific channel to generate a first task-specific feature map; generate an output for the first task-specific branch based on performing the shared function on the shared features and performing the first task-specific function on the first task-specific features; and update at least one parameter of the first gating mechanism based on the output.
Need to check novelty before this filing date? Find Prior Art