Even if one includes hidden download-sites and special access by university subscriptions, only sources at the low or medium levels are available in a sufficient amount.
How advanced are you referring to? There are many resources at the advanced undergraduate or beginning graduate level (this is biased towards computer science and math, so maybe this is not true for the sciences):
To get to the research level you certainly need journal access, which, as far as I know, is pricy for an individual, hence unrealistic. Hopefully more journals switch to open acccess.
In mathematics, I would call “advanced” roughly what is above the average of “Springer Graduate Text” book series level. Of course, I do not say that one finds nothing, e.g. the Bourbaki seminar series is very good for autodidacts, or this site. It is a bit like the difference between a complex organism and the bunch of isolated cells.
How advanced are you referring to? There are many resources at the advanced undergraduate or beginning graduate level (this is biased towards computer science and math, so maybe this is not true for the sciences):
This is a link to a freely available 4-volume set on measure theory
Here is a book on information-theoretic approaches to machine learning and inference
Here is one on statistical learning theory
Here is one on a “computational” approach to classical mechanics (Hamiltonians and Lagrangians abound)
To get to the research level you certainly need journal access, which, as far as I know, is pricy for an individual, hence unrealistic. Hopefully more journals switch to open acccess.
In mathematics, I would call “advanced” roughly what is above the average of “Springer Graduate Text” book series level. Of course, I do not say that one finds nothing, e.g. the Bourbaki seminar series is very good for autodidacts, or this site. It is a bit like the difference between a complex organism and the bunch of isolated cells.