Main Issues

[slerman12]

The main issue of Opposing Views is readability and believability, but I think the main content is there. I'm not convinced people have the attention span to read and appreciate all of this.

"Faith" is barely even drafted.

All of the parts related to my work during the program understate how much and what I've done, including the purpose and importance. I also still have to document UnifiedML, to make it possible for people to see for themselves, and for people in the field to see why it's worth using and better than any existing alternatives, or at least far and away can be.

[slerman12]

Work includes UnifiedML, generalists, Experience Replay Memory and DataLoader, Bittle, and XRDs.

[slerman12]

Purpose of PhD work, primarily: To democratize and organize in as un-convoluted a way as possible deep learning. This is independent from my post-PhD work of using my deep learning toolboxes for physics, which itself is an independent use from my during-PhD work on XRD pattern prediction, since the latter is about training predictive black-box models from data to the XRD patterns via labels down, and the former is about actual physics theory modeling from the quantum and relativistic levels up.

[slerman12]

Also to build long-term memory, but that got delayed.

[slerman12]

Much of this PhD-work purpose was generated incidentally, by the diverse array of projects I was dismissively assigned to (whereas my own purpose from early on was to build long-term neural memory and study intelligence/consciousness), and my own principled effort to unify all of them, not particularly supported or appreciated by the University of Rochester.

[slerman12]

Going so far as to terminate me for their poorly-judged infeasibility of that unification that I'd done.

[slerman12]

In addition to documenting, I have to wrap up that unification (pretty much as is, minor bug fixes) and write up the rest of my AI-Plan (some good parts of it are visualized in my open-sourced slide deck). My journal blueprints include an extremely elaborate dataset-organizing API (also partially found in comments of code), unifications for memory, more elaborate multi-modal (input and output), diffusion, and more elaborate self-supervised. I have to also write up some of the specific applications of deep learning more generally that I've blueprinted (some more recently than termination), and elaboration on the structure of the political and social philosophy (I've) summarized on animal-tree. I also came up with many algorithms that would be great to delegate to a research team.

[slerman12]

Perhaps another “main issue” is that I have to write my whole thesis on these subjects without being paid, employed, or supported.

Work that benefits everybody, work that people across fields want, work that others are getting paid big salaries or receive huge grants to do, work that others are wasting huge resources and money and manpower to figure out that I've already figured out, work that requires faith in me and faith in my ability to fully lead it even without a PhD, a lot of the implementation specifics of which are largely not the best use of my knowledge and skill: completely unpaid, currently unappreciated, the opposite of supported by the University of Rochester, whose jobs it was — if the purposes of those jobs are relevant to those jobs — to support my work, and with extreme respect.