It is clear that Anders’ presentation on definition of the sandwich was the big hit of the week and takes the prize for public admiration. I also admire the Anders’ research in this problem, as a lot of time has been spent arguing the “sandwichness” and Anders provides us with a rigorous topological basis for building such arguments. I also really enjoyed Joe’s and Bryan’s presentations. Applying TDA to topics that historically lie in the realm of humanities is very interesting. In particular, Joe’s presentation has involved some in-depth research in comparison of the newer methods to the methods of NLP analysis that are already used in industry. It is very good to know what works and what doesn’t, and Joe’s research provides analysis of TDA methods in such application.
April 29, 2018
Presentations Week 3
4 Comments
gljohns3
April 29, 2018 — 21:51
I was thinking about Ander’s definition for a sandwhich. and had an interesting chain of thought, but it abuses his definition of bread slices a little. A pizza with pepperonis replaced with little bread slices is then a sandwich (for wackos!). Now shrink the pepperonis and add more, then repeat the process indefinitely. Since nothing actually touches in nature, in the limit the little bread slices essentially form one bread slice connected to the crust (this is the abuse). So, strombolies are sandwiches. Then pop tarts are sandwiches, and we’ve now descended into madness. Even fruits with skin like tomatoes would then be sandwiches.
Seriously though, I enjoyed his talk, and I thought it was a neat point that vague definitions of things give vague instructions on how to make them.
jhmuelle
April 29, 2018 — 18:01
I agree with some bias that the presentations on TDA were very interesting. Bryan’s in particular was really cool. I hope more research gets done in that area. Studying music with math is just cool.
sroheda
April 29, 2018 — 15:46
I think Joe’s talk was indeed very interesting, especially since he also tried to compare his approach with neural networks usually used in machine learning. It may even be possible to combine the two approaches, where you could possibly use TDA to extract certain features from the data, and then feed these features into a neural network.
Nick
April 29, 2018 — 13:26
Joe’s and Bryan’s talk on TDA were really interesting to me. I hadn’t realised that topology could be used to analyze anything like music compositions or works of literature. I’m actually very interested to see how the vector space of words is defined so that you can do the word arithmetic Joe mentioned. That by itself seems like a fun tool to play with.