Sort:
Open Access Issue
Ensemble Knowledge Distillation for Federated Semi-Supervised Image Classification
Tsinghua Science and Technology 2025, 30(1): 112-123
Published: 11 September 2024
Abstract PDF (6.3 MB) Collect
Downloads:169

Federated learning is an emerging privacy-preserving distributed learning paradigm, in which many clients collaboratively train a shared global model under the orchestration of a remote server. Most current works on federated learning have focused on fully supervised learning settings, assuming that all the data are annotated with ground-truth labels. However, this work considers a more realistic and challenging setting, Federated Semi-Supervised Learning (FSSL), where clients have a large amount of unlabeled data and only the server hosts a small number of labeled samples. How to reasonably utilize the server-side labeled data and the client-side unlabeled data is the core challenge in this setting. In this paper, we propose a new FSSL algorithm for image classification based on consistency regularization and ensemble knowledge distillation, called EKDFSSL. Our algorithm uses the global model as the teacher in consistency regularization methods to enhance both the accuracy and stability of client-side unsupervised learning on unlabeled data. Besides, we introduce an additional ensemble knowledge distillation loss to mitigate model overfitting during server-side retraining on labeled data. Extensive experiments on several image classification datasets show that our EKDFSSL outperforms current baseline methods.

Open Access Issue
Deep Model Compression for Mobile Platforms: A Survey
Tsinghua Science and Technology 2019, 24(6): 677-693
Published: 05 December 2019
Abstract PDF (7.9 MB) Collect
Downloads:155

Despite the rapid development of mobile and embedded hardware, directly executing computation-expensive and storage-intensive deep learning algorithms on these devices’ local side remains constrained for sensory data analysis. In this paper, we first summarize the layer compression techniques for the state-of-the-art deep learning model from three categories: weight factorization and pruning, convolution decomposition, and special layer architecture designing. For each category of layer compression techniques, we quantify their storage and computation tunable by layer compression techniques and discuss their practical challenges and possible improvements. Then, we implement Android projects using TensorFlow Mobile to test these 10 compression methods and compare their practical performances in terms of accuracy, parameter size, intermediate feature size, computation, processing latency, and energy consumption. To further discuss their advantages and bottlenecks, we test their performance over four standard recognition tasks on six resource-constrained Android smartphones. Finally, we survey two types of run-time Neural Network (NN) compression techniques which are orthogonal with the layer compression techniques, run-time resource management and cost optimization with special NN architecture, which are orthogonal with the layer compression techniques.

Open Access Issue
SmartCare: Energy-Efficient Long-Term Physical Activity Tracking Using Smartphones
Tsinghua Science and Technology 2015, 20(4): 348-363
Published: 03 August 2015
Abstract PDF (17.2 MB) Collect
Downloads:132

Lack of physical activity is becoming a killer of our healthy life. As a solution for this negative impact, we propose SmartCare to help users to set up a healthy physical activity habit. SmartCare can monitor a user’s activities over a long time, and then provide activity quality assessment and suggestion. SmartCare consists of three parts, activity recognition, energy saving, and health feedback. Activity recognition can recognize nine kinds of daily activities. A hybrid classifier that uses less power and memory with satisfactory accuracy was designed and implemented by utilizing the periodicity of target activity. In addition, a learning-based energy saver was introduced to reduce energy consumption by adjusting sampling rates and the set of features adaptively. Based on the type and duration of the activity recorded, health feedback in terms of the calorie burned was given. The system could provide quantitative activity quality assessment and recommend future physical activity plans. Through extensive real-life testing, the system is shown to achieve an average recognition accuracy of 98.0% with a minimized energy expenditure.

Total 3