Viewing profile — togelius

We recently published a paper on increasing generality in machine learning through procedural content generation: https://www.nature.com/articles/s42256-020-0208-z if (paywall) the…

I had to make a few simplifications to spell out the differences clearly and avoid making the text infinitely long. It's true that most current gradient descent algorithms are stoc…

Use Google Scholar. Most technical researchers these days put all their papers on their webpage for free, and these days also on ArXiv.

Good point, I could have used "hacker" instead, though that's more specific to writing software. I didn't perceive "tinkerer" as a word with negative connotations when I wrote it. …

Oh yes, I'm a terrible hacker, I rarely write code anymore. An unfortunate byproduct of academia is that you get "promoted out of the job", and all the hands-on work is done by you…

"a systematic enterprise that builds and organizes knowledge" That is exactly my point. Doing your scholarship is doing your part to organize knowledge. Systematically building kno…

I'm the author, and I'm somewhat amused by the negative tone of many of the comments. Seems there are plenty of people with resentments towards research here ;)

This is all true. However, the large publishers (IEEE, ACM, Springer) these days all allow you to self-archive on your own webpage and typically also on ArXiv. So you can (and shou…

He mentions NEAT, which is good, but not any of the work on playing Mario, much less any work on evolving neural nets to play Mario. There's quite a few papers on the topic, not ju…

Like basically all researchers in CS these days, I keep all my papers freely downloadable from my webpage. Easily findable via Google Scholar, or just a regular Google search, for …