Caltech Stem

This week, our group split into two groups: those that are working on the same dataset, and the other that is organizing a new dataset. I am in the first group and I finally decided that I would be investigating graph clustering with missing data. The other group is building a smaller dataset of fruits. There are two categories: strawberries, and wedge strawberries. Fifteen pictures from each category are included in the small dataset. Below is a picture of each. In regards to my contribution, given a graph with partial observations, I still want to identify the underlying clusters. This leads me to the problem of clustering graphs with missing data. For the next few weeks, I will consider the problem of identifying clusters when the input is a partially observed adjacency matrix of an unweighted graph. Two convex algorithms will be my aim but are not completely promised. A video about this topic is below: For the other group, I do not know the specifics but I understand whe

On Thursday, we met with Dr. Hassibi to talk about what we have been doing after last time. What I did before our meeting was trying to tinker with the eigenvalues and eigenvectors. In my attempt to do Kmeans on the dataset, I stored the first k-number of eigenvectors and vertically stacked them on to each other into another column matrix. I then ran Kmeans on it but that was not the correct execution as pointed out by Dr. Hassibi. He said to take the eigenvectors and put them next to each other (stacking horizontally) and then run it. It makes sense since stacking vertically would mean there would be k*n # of nodes. Here is the photo when we plotted our results and below is a video of our meeting. A graph showing Kmeans clustering nodes into three classes. As for the future, we were given two options: keep experimenting with the dataset at hand or try out a new one. With the current data, much is there to be explored such as test the question of whether a pair-wise or tri