Yes, very nice presentation. I came on this topic because I tried to learn more about machine learning and the free energy principle was mentioned. It is very hard to find resources on the internet that explain it without math. This presentation also has some math in it but at least it is a coherent story that allows me to form a picture about the main idea. And I guess I now have to minimize my observation errors by getting the math in my head 🤔 or by convincing other people that this needs more explanation without math 😅
Loved the introduction, problem layout and philosophical undertones highlighting the wide scope of these concepts. The last bit got a little messy which is difficult to avoid when sharing some of the mathematical foundations. That section would in my opinion benefit from some examples that would set the model in a practical context. Otherwise, one of the best presentations of the topic I've seen. Thanks for uploading.
scientists change their models, artists change the world :P To develop a little my catchy motto: to minimize their PE, scientists (are known to mostly) change their models, aritsts (are known to mostly change) their worlds. That may explain the crazy and active behaviours of artists and the passive and thoughtful behaviours of scientists. Dont hate me, i know this is still too general a model of them, but it's just a little relative map. If i give a full definition to minimize my error, it wouldn't be that handy or catchy (= easily digested and compatible with everyone),. That now says a lot about the mottos, slogans marketing and news in general but also their receivers' predictions. i will avoid getting political. I love how this perspective makes me live in the meta :P
I think this is pretty accurate in the sense that we ourselves embody the logic of the system: there are modelers and there are doers. The doers focus on getting things done according to whatever model is currently the status quo, while modelers are constantly checking new models as to better fit the ever-changing incoming data. That being said, I think we could all learn to lean a bit the other way (doers could think a bit more, and thinkers could sometimes engage in some doing!)
