FieldSHIFTing for the SoftComp Intelligent Soft Matter Workshop

All right. Hello and welcome. It is August 5th, 2024. This is Active InfraAnt Stream, 001.1, new stream series just dropped. And this is where we're headed today. We're going to be doing linguistic analysis on fully synthetic all-by-all dissertations. We're going to go into a lot more detail on what that means, but just quickly at the top level, each of these points in a principal component semantic space is a written dissertation shifting from one field to another, following on on the lab of Mike Levin et al., who worked on the field shift one. This is field shift two. It does quite a few different things. And for example, this data point is the actual term frequency usage distribution on synthetic negotiation to synthetic quantum computation. I'll talk more about what the synthetic part means. This is like a PhD dissertation that's going from negotiation to quantum computation. And here we have similarity clustering across in all-by-all of different domains. So we're going to be going to a lot of different places and ways. Let's just start with this text file off the bat. Super excited for this. It's been really rapidly developing. I'm using cursor 0.39 came out just a few days ago, even better than the prior ones. So let's start this off, right? I'm going to push 3000 files to a major Git release. This is going to take a few minutes. Once it pushes, push that now. All right. Once that pushes, then I'm going to come over to Zenodo and make a new release on GitHub. So first just wanted, while that's loading, you can see it on the top of the page. Zenodo makes it super easy to connect GitHub repos via Zenodo who offers DOIs, digital object identifiers. So active inference is right here. Once the GitHub push is completed, then we'll make a release of the repo. And then that will get a new DOI and archive it on Zenodo. So that's really awesome. So just a few different tools using Zenodo for publishing for DOI, GitHub for a code. And let's reload it, see if we're in. One minute ago, active inference stream 001. So now let's make a new release. This is going to be number 01.1. Needs a tag. 1.01. During the stream. Publish it. 1.1 is published. Reload on Zenodo. And it'll probably take a few minutes, but that's pretty cool because then it'll also have the DOI coming from Zenodo and the archiving. So maybe we'll check back there later. Okay. Let us go to... These are... It's in the background. It's translating the dissertations into Sanskrit and Arabic in the background. So hence these new files being added, but we're going to get all there. Okay. Start with the GitHub push. Done. Reload and confirm. Update. Make the GitHub update. We did that. And we're using cursor 0.39. Came out a few days ago. Okay. For backup. There's a page in the active block for its the platform. So it summarizes a little bit about field shift. So I'm doing this. That's right now. August 5th, 16 UTC. The code's in active inference. Just uploaded now. What field shift 2 is doing right now is generating thousands of these all-by-all domain A to domain B. And I borrow this term and develop in the same tradition from Mike Levin et al.'s work. So here's the paper here. Machine learning for hypothesis generation in biology and medicine, exploring the latent space of neuroscience and developmental bioelectricity. O'Brien et al. Levin. So they briefly... Let's see if the figures look bigger. They briefly describe how AI is playing an important and pervasive role in the scientific process. They talk about analogies and isomorphisms between, for example, the neural synapse and the cellular junction, bioelectrically and chemically considered. They talk about some even broader, quote, functional invariance between neuroscience and developmental biology. So that sounds kind of cool, promising. Of course, metaphor. Here's the BART-based domain translator. This relates to taking material from one content and then putting it into an encoder. It gets decoded. And here they provide…