Authors: D Rawlings, T Chopard
Summary: This research focuses on developing a multi-label classification method for predicting plant species using environmental factors like satellite images and climate data. It employs an ensemble of machine learning models to handle multi-modal data and variable label counts, contributing to biodiversity conservation efforts.
Authors: T Chopard, D Rawlings
Summary: This paper presents a generalizable BERT-based approach for identifying and classifying sexism across tweets, memes, and videos. By training on text and applying models to other media via OCR and annotations, the study demonstrates that a single model can effectively handle multiple media types with minimal preprocessing.
Authors: T Chopard, D Rawlings
Summary: This study investigates a two-step deep learning approach to enhance Species Distribution Modelling by combining Presence-Only (PO) and Presence-Absence (PA) data. Pre-training on large PO datasets before fine-tuning on PA data significantly improved prediction accuracy, demonstrating the value of integrating diverse data types for conservation.