- Most computer science research can be categorized into one of three forms: preprints published on arxiv or similar websites, papers published in conference proceedings, and source code. In artificial intelligence research, it’s typical for research, if there’s demand, to be promoted from a preprint, to a paper published in a conference proceeding, to publicized source code. Most research doesn’t make it past the first step and that’s usually okay.
- Many computer science labs are open to collaboration from outside contributors. It’s especially true for academic labs where day-to-day research is usually led by graduate students and post-doctoral students. In either situation, regardless of domicile, that person is underpaid and overworked and will eagerly accept help. However, a lot of computer science research comes from for-profit institutions (e.g. Facebook AI Research, Google Brain, and Microsoft Research are three well-know institutions in artificial intelligence research). In my experience, labs inside for-profit institutions will readily work for academic and non-profit institutions for the prestige but are less likely to work with independent contributors since there’s less incentive.
- When you start looking for contributors, I’d look for academic or non-profit labs that match your specific interest, I’d look for the people in that lab that are actually publishing (i.e. check arxiv), and I’d directly offer your skillset from the start (e.g. “I’m a programmer that’s worked on such-and-such and I’m looking to contribute”).
Finally, if you working on computer vision, please reach out! My lab at the Broad Institute of MIT and Harvard are always looking for free help and more than willing to help people bootstrap their publication record.
Well, that and lawyers. Even getting academic collaborations approved gets super complicated once lawyers are concerned about patents, etc.
It has the potential to significantly reduce complexity in various areas of the signal processing stack (in exchange for greater software complexity and black boxes)
We plan on using it to process signals for LED-based satellite FSO systems.
Would you like to get in touch to discuss more about it? We could use some advice on software for atmospheric simulations.