Hi, I am Amelia, a fourth year Ph.D. candidate at the BCN MedTech group, under the supervision of Gemma Piella and Diana Mateus. My research interests are in the broad areas of medical imaging and deep learning.
I work on the design of data schedulers, i.e. a deep learning algorithm which determines the order and pace of instances presented to the optimizer, aiming to improve classification accuracy.
Specifically, I focus on curriculum-based strategies to develop data schedulers able to deal with class-imbalance, unreliable or limited amounts of annotations and domain shift, problems that are fairly common in medical datasets.
- October 14th: I have successfully defended my thesis "Learning Representations for Medical Image Diagnosis: Impact of Curriculum Training and Architectural Design".
- December 2020: "Curriculum Learning to Deal with Noisy Labels" accepted for oral presentation at iTWIST'20.
- August 2020: I have attended the Deep Learning & Reinforcement Learning Summer School (DLRLSS 2020).
- June 2020: "Precise Proximal Femur Fracture Classification for Interactive Training and Surgical Planning" has been accepted for oral presentation at IPCAI 2020 and journal publication in IJCARS.
- June 2020: "Hierarchical Deep Curriculum Learning for the Classification of Proximal Femur Fractures" has been accepted for oral presentation at CARS 2020.
- Dec. 2019: Poster presentation at the Deep Learning Barcelona Symposium (DLBCN 2019).
- Oct. 2019: Poster presentation of "Medical-based Deep Curriculum Learning for Improved Fracture Classification" at MICAAI 2019 in Shenzhen, China.