Address:
Aymane Abdali frome the mathematical and electrical engineering départment (MEE) and LABSTICC laboratory, wil present his research about :
"Few shots learning : data selection, performance prediction and applications"
Abstract : Machine Learning continuously brings numerous benefits in industries where AI-powered tools are optimizing manufacturing processes, predicting machinery failures, enhancing quality control with real-time defect detection, etc. However, numerous challenges persist in addressing the diverse requirements of real-world applications. This thesis focuses on one such challenge, specifically related to data efficiency and learning in Few-Shot settings. The research presented here investigates methods to enhance data efficiency by leveraging human knowledge and expertise. This exploration extends beyond merely acquiring and labeling additional data to include approaches such as refining existing annotations and utilizing tools to optimize the annotation process
Organizer(s)
Thesis co-acreditation from IMT Atlantique with the doctoral School SPIN
Keywords : Few-Shot, Image Classification, Active Learning