Umm…How did I join Swarma?

I joined the swarma club when I saw Wocheng Fu’s ADS in zhihu (the Chinese answer to Quora) and thought the workshop’s topic is quite close to what I have always wanted to do: use machine learning to study physics and perhaps. to be specific, something about tensor netowrk, quantum computing and macine learning. So I sent an email and found that I was accepted (I thought it seems impossible to be accepted since I was an undergraduate…)

And the place where we had this workshop is quite amazing! I thought it would be in a caffe in Beijing. But the Kaifeng foundation supported us to hold it in Beijing water town, a pretty beautiful place near Beijing…

So what’s it about?

In my opinion, we mainly discussed Complex Network, Hyperbolic Geometry, Quantum Mechanics, Statistical Mechanics (Perhaps mainly about the information part, like belief propagation), Machine Learning Algorithms (mainly about Boltzmann machine and GAN). I did neither wide nor deep knowledge about all those things at first, but I have to say, dude, I really learnt a lot in those days. And Yizhuang You’s talk is quite amazing: a Boltzmann machine from random tensor network, which means we could use a the training process of a fully connected Boltzmann machine to explain how the geometric structure was formed in our universe! And this Boltzmann machine I guess, should be something like the Quantum Boltzmann Machine from an arixv paper this year. (A Dwave’s work?). But, well, a fully connected Boltzmann machine is really hard to train… But what if we use a quantum computer? I have not considered the complexity of a quantum Boltzmann machine and random tensor network.

Another interesting one is GAN, but I need some time to figure out the materials. It sounds intersting: you use a machine to guide a machine to learn! I’ll post a note about GAN later…

And the ISOMAP, I heard this algorithm from some machine learning guys before (about geometric learning), but I did not read the theory… learnt this from Lingfei Wu

Well, I still need prepare for PhD application this year, and since my summer and spring was messed up due to some personal problem, it becomes more and more busy recently…So I guess new posts about those theories and algorithms will come a little bit late…

Photos in water town

Though, talks and discussions were from 8:30 in the morning to 5:30 in the evening, we still had some time to hang out in water town, so there are some photos. It feels really good to discuss questions in such a beautiful and clean place so that I and symcheng were like random walk in the town XD