Task-Adaptive Incremental Learning for Intelligent Edge Devices

TL;DR

This paper introduces TeAM, a task-adaptive incremental learning framework that enables CNNs on edge devices to be efficiently customized and collaboratively improved with local data, addressing redundancy and dynamic data challenges.

Contribution

TeAM is a novel framework that allows pre-trained CNNs to be quickly adapted for specific edge tasks and collaboratively aggregated for improved global performance.

Findings

TeAM achieves high accuracy with quick fine-tuning on local data.

Collaborative aggregation enhances global model performance.

Efficient handling of dynamic data in edge applications.

Abstract

Convolutional Neural Networks (CNNs) are used for a wide range of image-related tasks such as image classification and object detection. However, a large pre-trained CNN model contains a lot of redundancy considering the task-specific edge applications. Also, the statically pre-trained model could not efficiently handle the dynamic data in the real-world application. The CNN training data and their labels are collected in an incremental manner. To tackle the above two challenges, we proposed TeAM a task-adaptive incremental learning framework for CNNs in intelligent edge devices. Given a pre-trained large model, TeAM can configure it into any specialized model for dedicated edge applications. The specialized model can be quickly fine-tuned with local data to achieve very high accuracy. Also, with our global aggregation and incremental learning scheme, the specialized CNN models can be collaboratively aggregated to an enhanced global model with new training data.