CIRCL: Jen Seale (Computational Linguistics)
DEC 04, 2018 | 6:30 PM TO 8:00 PM
The Graduate Center
365 Fifth Avenue
December 04, 2018: 6:30 PM-8:00 PM
Jen Seale presents on a topic in computational linguistics.
Bi-modal classification with attention and transfer learning:
A discussion focused on implementation
Recent work by Kiela et al. (2018) reveals that bi-modal image and text classification neural net models far outperform both text- and image-only models on media that incorporate both text and images (e.g., online blog posts, advertisements). My research will extend that work by determining if accuracy in classification may be increased by the implementation of features that mimic human attributes: attention implemented in a pre-trained, fine-tuned computer vision model, and transfer learning implemented for language processing. Performance over a multi-modal IMDB data set will be analyzed and compared to the baseline provided by Kiela et al. (2018). This talk will focus on the basics of implementation with PyTorch.
All are welcome!