When I left my previous lab, I was at least able to secure a written confirmation stating that I am free to pursue the idea I had proposed on using three-factor learning rules for training RNNs independently. While it might seem unusual to have to ask for permission to work on ideas you proposed and never worked on in the lab, in some situations it is simply prudent to have this documented. This discussion came after my former advisor had already ensured that competing labs which might have been open to collaborating on it with me would not do so. I suspect the assumption was that I would be unable to develop the work on my own because I would be occupied with other projects, given that my access to those labs had been blocked.
For context, I had originally been scheduled to co-supervise a Master’s student on this topic together with my former collaborator, Matt Getz. However, on 30th October 2024, I received an email informing me that this supervision would no longer take place, i.e. the student would get supervised only by Matt and not by both of us, but he would still be working on the idea I proposed.1
I am sharing the email here so that, when I present my work in the future (which is based entirely on my own methods and does not use any material from the related project in my previous lab), there is no room for misrepresentation regarding the origin of the ideas or the work.
For the record this is the last stage of the project I was inovled in in my previous lab on a similar topic, with two-factor rules and low-rank recurrent networks, presented at the Bernstein Conference 2024 is here: Bernstein 2024 poster .2 And here is the presentation mentioned in the email where I proposed to employ three-factor rules to actually learn a task in a bio-plausible way in RNNs presentation 3.
Footnotes
This came along with the invalidation of our written agreement (that my former postdoc advisor proposed to draft in the first place) on how to continue the three projects I was involved in, pretending a lack of records although we presented these three projects multiple times in conferences and she had final versions of posters, abstracts, my presentation slides etc. (At the end of my contract we presented all threee projects I was involved in at Bernstein Conference 2024, so my advisor was fully aware of the state of the projects. Moreover, my last day in the lab before leaving for Bernstein I gave a group meeting presentation that was quite extensive and lasted nearly two hours)
↩︎You might need to refresh several times to see the poster, but you can also use this link: my link↩︎
The presentation was a joint one together with my collegues Matt Getz and Pablo Crespo (first part presented by matt, second by me, and third by Pablo).↩︎
