I am Qiao, currently a PhD candidate at TU/e. In 2018, I got my master’s degree and joined in an AI company as an Algorithm Researcher. Traditionally, for the training of deep learning models, all data has to be transferred from the edge to the cloud, which creates unnecessary problems and challenges, from the engineering aspect of managing large amounts of data to privacy issues.
In the MegaMind program, I get the chance to extend deep learning processing capabilities from the cloud to the edge in electricity grid. The title of my research track is “Scalable Decentralized Learning Algorithms” with the aim of developing scalable (decentralized) learning algorithms for smart grid energy system, which makes it possible to reduce computational and memory requirements and solve privacy issues.
Another reason why I joined the MegaMind program is a large number of partners, including universities and companies as well, which makes it easy for me to know the practical problems in real-world applications and solve these problems in the near future.
During this PhD period, I hope to contribute something to the green and safe AI in the electricity grid. It will have a significant environmental impact on reducing the energy cost of the AI models and CO2 emission.
Publications
Dynamic Sparse Training versus Dense Training: The Unexpected Winner in Image Corruption Robustness
Wu, B., Xiao, Q., Wang, S., Strisciuglio, N., Pechenizkiy, M., Van Keulen, M., Mocanu, D.C., Mocanu, E. – 2025
E2ENet: Dynamic Sparse Feature Fusion for Accurate and Efficient 3D Medical Image Segmentation
Wu, B., Xiao, Q., Liu, S., Yin, L., Pechenizkiy, M., Mocanu, D.C., Van Keulen, M., Mocanu, E. – 2024
Are Sparse Neural Networks Better Hard Sample Learners
Xiao, Q., Wu, B., Yin, L., Gadzinski, C. N., Huang, T., Pechenizkiy, M., & Mocanu, D.C. – 2024
Dynamic Data Pruning for Automatic Speech Recognition
Xiao, Q., Ma, P., Fernandez-Lopez, A., Wu, B., Yin, L., Petridis, S., Pechenizkiy, M., Pantic, M., Mocanu, D.C., Liu, S. – 2024
Achieving Long-term Time Series Forecasting Models with Fewer Than 1k Parameters through Dynamic Sparse Training.
Xiao, Q., Wu, B., Pechenizkiy, M., & Mocanu, D.C. – 2024
Can Less Yield More? Insights into Truly Sparse Training
Xiao, Q., Wu, B., Yin, L., Van Keulen, M., Pechenizkiy, M. – 2023
Dynamic Sparse Network for Time Series Classification: Learning What to” see”
Xiao, Q., Wu, B., Zhang, Y., Liu, S., Pechenizkiy, M., Mocanu, E., Mocanu, D.C. – 2022