ChannelGate: A Gating Network to Route Ion-Channel Data towards Optimal Deep-Learning Models
Project Overview
ChannelGate automatically routes each ion-channel recording to the most suitable Deep-Channel model.
What were the key results of your research project?
ChannelGate achieved 96.7% file-level accuracy and a Macro F1-score of 0.964 on an independent test set of 60 files. The results showed that a lightweight CNN-based classifier could reliably distinguish between Deep-Channel model groups and recommend the most suitable model for each recording. The system was also successfully converted to TensorFlow.js for integration into the Deep-Channel web application.
GitHub Repository: https://github.com/wendyjiang-hub/channel-gate/tree/wendy-channelgate
A presentation of this research will be shared here and on our YouTube site when available.
How do you feel you have benefitted from completing this internship and has it made you consider future career paths?
This internship gave me a valuable opportunity to experience what research is like and helped me better understand what doing a PhD might involve. It encouraged me to think more seriously about my future study and career options.
I also learned how to use HPC resources to solve practical machine learning problems and gained experience working collaboratively with researchers and other team members. Overall, the internship improved both my technical skills and my understanding of research and teamwork.
Download presentation slides