4 ms·
After a quick glance, this sounds a lot like information theoretic empowerment, which has been around for more than five years: http://www.plosone.org/article/
by mahrz 13y ago
After a quick glance, this sounds a lot like information theoretic empowerment, which has been around for more than five years:
http://www.plosone.org/article/info%3Adoi%2F10.1371%2Fjournal.pone.0004018 http://www.plosone.org/article/info%3Adoi%2F10.1371%2Fjourna...
https://uhra.herts.ac.uk/dspace/bitstream/2299/6712/1/901919.pdf https://uhra.herts.ac.uk/dspace/bitstream/2299/6712/1/901919...
https://uhra.herts.ac.uk/dspace/bitstream/2299/9659/1/ContiPaper.pdf https://uhra.herts.ac.uk/dspace/bitstream/2299/9659/1/ContiP...
Also the concept of homeokinesis is quite similar and also not quoted in this work:
http://www.informatik.uni-leipzig.de/~der/Forschung/homeoki.html http://www.informatik.uni-leipzig.de/~der/Forschung/homeoki....
- irickt 13y agoThanks for posting these. The third link in particular applies a very similar approach to the inverted pendulum example, in depth and with clarity. I also note the modest concluding paragraph for comparison to the grand claims of the OP. "Finally, we conclude that it would be desirable to transfer some of the generality of the underlying empowerment concept into methods for empowerment calculation in systems combining continuous and discrete dynamics, especially systems of real-world relevance."