We introduce exBERT, a training method to extend BERT pre-trained models from a general domain to a new pre-trained model for a specific domain with a new additive vocabulary under constrained training resources (i.e., constrained computation and data). exBERT uses a small extension module to learn to adapt an augmenting embedding for the new domain in the context of the original BERT's embedding of a general vocabulary. The exBERT training method is novel in learning the new vocabulary and the extension module while keeping the weights of the original BERT model fixed, resulting in a substantial reduction in required training resources. We pre-train exBERT with biomedical articles from ClinicalKey and PubMed Central, and study its performance on biomedical downstream benchmark tasks using the MTL-Bioinformatics-2016 dataset. We demonstrate that exBERT consistently outperforms prior approaches when using limited corpus and pretraining computation resources.
We propose a race path synthesis framework based on a data-driven approach that provides good controllability for synthesizing race paths with characteristics preserved for racer animations. We introduce formation field, a data structure that samples regions in formation space that contains formations of exciting and realistic race paths, generated using a set of collected race paths in a path database. By traversing the regions according to a given constraint, we generate a path in formation space that defines how to synthesize the desired race path by interpolation. Because the new race path is synthesized from existing paths with quality guaranteed, it also provides the same level of quality. As the experimental and user study results show, our framework produces good results effectively and is suitable for both real-time applications such as horse racing games and race-path-generating tools.
Common methods for reducing image size include scaling and cropping. However, these two approaches have some quality problems for reduced images. In this paper, we propose an image reducing algorithm by separating the main objects and the background. First, we extract two feature maps, namely, an enhanced visual saliency map and an improved gradient map from an input image. After that, we integrate these two feature maps to an importance map. Finally, we generate the target image using the importance map. The proposed approach can obtain desired results for a wide range of images.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.