|
|
|
|
|
|
by leakydropout
3177 days ago
|
|
|
One I forgot is to work with either algorithms or datasets created by other researchers. Once I send an email asking for access to a certain dataset, describing the idea I had for it, and had a very famous researcher reply that they'd be interested in a cooperation. The other side of this advice is to create a dataset that is interesting to other researchers. |
|
|
I had a similar experience, as somebody who is also a "college dropout" and not formally associated with academia at all. I emailed a professor who wrote a book I was reading and asked for access to some of the datasets he cited in the book, and explained that I wanted to try re-implementing his technique using a newer tech stack (the book was from the 80's mind you) and then look at extending the ideas somewhat.
He quickly replied with the data, a pre-print of a new paper he was working on, and an invitation to keep him in the loop on my work. Not an outright invitation to collaborate, but I suspect if I achieve a useful result, the opportunity may well be there.