how is the state of the arts in transfer learning and semi-supervised or active learning? which one would you choose as a PhD topic and why? I have the possibility to pursue a PhD in one of these topics. I'm confused which one to choose. I took a look at the actual research in both fields and it looks promising. I have also interests in both fields that's why I'm having a hard time to decide since I must choose something concrete for my DIS topic.
It purely depends on the problem you have chosen. Transfer learning give better results. Since active learning has to be done from the scratch so the results are on lower side.
The recent SOTA like Yolo, BERT etc are used by researchers by following transfer learning approach on different problems. Even GPT is based on transfer learning and it is the probable reason behind its success
- Badges
- Science topic
- Similar topics
- Correction