p-Meta: Towards On-device Deep Model Adaptation
This addresses the need for memory-efficient adaptation in IoT devices with private and diverse data, representing an incremental improvement over existing gradient-based meta learning schemes.
The paper tackles the problem of on-device deep model adaptation for IoT applications, which requires data and memory efficiency, by proposing p-Meta, a meta learning method that enforces partial parameter updates. The result shows that p-Meta improves accuracy and reduces peak dynamic memory by a factor of 2.5 on average compared to state-of-the-art methods in few-shot tasks.
Data collected by IoT devices are often private and have a large diversity across users. Therefore, learning requires pre-training a model with available representative data samples, deploying the pre-trained model on IoT devices, and adapting the deployed model on the device with local data. Such an on-device adaption for deep learning empowered applications demands data and memory efficiency. However, existing gradient-based meta learning schemes fail to support memory-efficient adaptation. To this end, we propose p-Meta, a new meta learning method that enforces structure-wise partial parameter updates while ensuring fast generalization to unseen tasks. Evaluations on few-shot image classification and reinforcement learning tasks show that p-Meta not only improves the accuracy but also substantially reduces the peak dynamic memory by a factor of 2.5 on average compared to state-of-the-art few-shot adaptation methods.