Advice on dissertation project s.t. it could be relevant to OpenCog

[Maria Kapros]

Hello everyone,

I am Maria Kapros, 4th-year Computer Science student at University College London. Have been following the OpenCog work for a year now and figured that it would be good to start contributing.

I am writing this message to ask for suggestions regarding the following. I am thinking of choosing my dissertation project such that it has some relevance to the AGI field and ideally OpenCog if I get my supervisor's approval. 

My idea is to work on improving ML algorithms for automated theorem proving, along the lines of https://arxiv.org/pdf/1905.10501.pdf. I came across this old forum (https://community.singularitynet.io/t/automated-theorem-proving-and-agi/255) saying that an improved theorem prover would be useful if integrated with URE. Is this still an active task that could be picked up? 

If not, what would be a good (challenging from both theoretical and implementation p.o.v) starting idea for a dissertation project that would be relevant to the OpenCog community? My supervisor's research interests are data-efficient learning, reinforcement learning, and meta-learning, so I think my previous questions will concern OpenCog's high-level components (perhaps MOSES) rather than atomspace.

Many thanks for reading this long message and please let me know what are your thoughts regarding my questions.

Best,

Maria