This critique will be based on two papers listed below on unsupervised graph representation learning. Your critique should be no more than 1500 words. Please submit your critique as a PDF via Canvas. The papers are:
- node2vec: Scalable Feature Learning for Networks
- OhmNet: Predicting multicellular function through multi-layer tissue networks
Your critique should cover each of the following points. Please dedicate at least one section to each of these points.
- What is the general problem area these papers are addressing?
- Briefly describe the motivation of the method described in each paper and how the method accomplishes its designated goal.
- What novel results and insights were obtained using these methods?
- Both papers are based on the node2vec algorithm. Compared to other unsupervised representation learning approaches on graphs, what are some of the key strengths and weaknesses of node2vec?