AI CUP - 2nd Place

As a single-person team in the AI CUP 2019, I earned 2nd place out of 469 teams. This competition centered on developing a system to automatically read and classify sentences in computer science paper abstracts into categories like Background, Objectives, Methods, Results, Conclusions, or Others, where sentences could have multiple labels. This task required me to build and train effective natural language models capable of understanding the nuanced language of academic research.

To achieve this, I utilized Tensorflow 1.15 and applied transfer learning techniques with prominent models such as BERT, SciBERT, BioBERT, and XLNet. This approach enabled me to develop a highly accurate system for classifying sentences from a substantial dataset of 47,000 abstracts. My success in this competition showcases my proficiency in natural language processing, machine learning, and the practical application of deep learning frameworks and advanced transformer models to solve complex text analysis challenges.