Google Developer Student Club, University of Malaya
Inspired by transfer learning successes in natural language processing (NLP), I shared about developing a deep learning model that is applicable for protein characterization tasks including protein family classification and protein-protein interaction prediction.
Pacific Northwest Quantitative Biology Symposium
Discussed controllability of networks, structural controllability and how the low controllability of the genetic regulatory network of E. coli can provide biological insight.
St. Stephen's College Feynman Club Colloquium
Discussed data collection and the use of sentiment analysis with recurrent neural networks to classify tweets as pro or anti terrorist. This was followed by a presentation of the network structure of a particular network of pro-terrorist Twitter users. Concepts such as scale-free networks and robustness of networks were introduced in this context. The talk concluded by considering the relationship between the qualitative nature of the tweets and the topological features of the network.