I am on the 2023-2024 job market. Feel free to check my Resume and contact me at blia@dtu.dk.
I am a final-year PhD student (expected to graduate in Jan-2024) at the Technical University of Denmark (Department of Applied Mathematics and Computer Science), and an affiliated researcher at the AI Pioneer Center, under the supervision of Tommy S. Alstrøm and Mikkel N. Schmidt. I have also been fortunate enough to be supervised by Sebastian U. Stich during my external research stay at CISPA Helmholtz Center for Information Security, Germany.
Before my PhD, I obtained my BSc in Financial Engineering at Jilin University (2012-2016), my MSc in Mathematical Modelling and Computation at the Technical University of Denmark (2016-2018) with a scholarship from DABAI, and worked as a researcher (2019-2021) under the supervision of Sam Leroux and Pieter Simones at Ghent University - imec.
My research is mainly about developing federated machine learning tools for chemical hazard detection using Raman spectrum and Surface Enhanced Raman Spectra (SERS). Specifically, I am focusing on developing machine learning tools for bacterial identification [Analyst, 2022] and explosives detection [Analyst, 2023], investigating non-convex federated learning from the optimization perspective [CVPR, 2023] [Preprint], and uncertainty quantification [ICML workshop, 2022]. Previously, I also did research in detecting anomalies in surveillance videos under adverse weather [CVIU, 2021][WACV, 2022] and developing distillation learning-based hardware-efficient framework for traffic counting [CEE, 2022].
I am interested in federated learning, representation learning, optimization, uncertainty estimation and more.
Selected publications, work experience, and projects supervision
An improved analysis of clipping in federated learning (2024, ICLR)
BL*, Xiaowen Jiang*, Mikkel Schmidt, Tommy Alstrøm, Sebastian Stich, Paper
01/2024Deep learning project supervision
Investigate the quality of the estimated uncertainty from using federated learning (four groups)
10/2023-12/2023Synthetic data shuffling accelerates the convergence of federated learning under data heterogeneity (2024, TMLR)
BL, Yasin Esfandiari, Mikkel Schmidt, Tommy Alstrøm, Sebastian Stich, Paper, Code
10/2023Nitroaromatic explosives detection and quantification using attention-based transformer on SERS maps (2023 Analyst)
BL, Giulia Zappalà, Elodie Dumont, Anja Boisen, Tomas Rindzevicius, Mikkel Schmidt, Tommy Alstrøm, Paper, Code
08/2023Bachelor thesis supervision
Exploring representation learning in federated learning. Victor T. Olesen and Rasmus S. Mikkelsen
01/2023-05/2023Partial Variance Reduction improves Non-Convex Federated learning on heterogeneous data (2023, CVPR)
BL, Mikkel Schmidt, Tommy Alstrøm, Sebastian Stich, Paper, Code
09/2022-01/2023Project supervision
20 RASPBERRY PI’S, one model: Federated learning on hardwares Søren W. Holm (now at Alvenir), Asger L. Schultz, Gustav L. Moesmand (now at Alvenir)
02/2022-06/2022Raman Spectrum Matching with Contrastive Representation Learning, 2022 Analyst
BL, Mikkel Schmidt, Tommy Alstrøm, Paper, Code
04/2022Project supervision
Unsupervised representation learning. Till A. Aczel (now PhD student at ETHZ)
10/2021-12/2021Master thesis supervision (coding support)
Segmentation of cardiac structures based on MRI data. Katrine M. Ejlev (now at ChemoMetec) and Michala Z. Blicher (now at 3Shape)
03/2021Decoupled appearance and motion learning for efficient anomaly detection in surveillance video (2021, CVIU)
BL, Sam Leroux, Pieter Simoens, Paper, Code
2021
Researcher at Ghent University-Imec, Belgium
Developing hardware-efficient deep neural network for detecting anomalies and counting traffic in surveillance videos
01/2019-12/2020Research assistant at DTU compute, Denmark
Developing uncertainty estimation based active learning framework using region acquisition strategy for medical image segmentation
09/2018-11/2018Skills
Python
Rstudio
Matlab
Bash script
Linux
PyTorch
Tensorflow
AWS
Github
React
Languages
Native language
Professional
A1, 365+ duolingo streak