• Applied transfer learning using ResNet34 for the purpose of thoracic disease classification, using multi-label chest X-rays • Trained/compared model performance using random vs. pretrained weight initialization and applied data imbalance considerations to assist with model performance