Exploration of BERT-BiLSTM models with Layer Aggregation (attention-based and capsule-routing-based) and Hidden-State Aggregation (attention-based and capsule-routing-based).

We design a multi-task model for joint disaster classification and victim detection. We train the model using both the Centralized Learning (CL) and Federated Learning (FL) methods. We also tried Active Learning (AL) to see how it could help in reducing the labeling workload for disaster dataset. Lastly, we optimized the model using OpenVINO.

An image processing model to classify the type of disaster based on the image dataset.

Train a lightweight MobileNetV2 model for disaster classification using the Crisis Image Benchmark Datasets (CrisisIBD).