Alex Mikhalev

How To Get Into Independent Research On Alignment/Agency

Thank you for this post very encouraging - I was thinking about applying to LTFF - I have all pre-requisites, now I feel it’s worth the try.

What's So Bad About Ad-Hoc Mathematical Definitions?

"partial credit" for getting *close* to the right answer" - this is where probability theory comes into play: "What is the likelihood of the enemy reading the message if it's intercepted in fullness".

Also, the concept which helps me to think about how to bring abstract concepts into reality is the time or any notion of temporal dependencies, which probability theory missed at large for a while and now recovering: see nature article.

Rage Against The MOOChine

I would recommend structuring applied maths learning differently: start with Computational Beauty of Nature https://mitpress.mit.edu/books/computational-beauty-nature and then go deep in relevant areas + graph theory + graph algebra. Also a deep understanding of multi-objective optimisation techniques: NSGA-3, Pareto front/Pareto surface.

Rage Against The MOOChine

I agree that nothing bets practical projects, but in modern life, you need to learn a lot of background information before jumping into the real world. There are plenty of ML projects and examples that are equivalent to the ToDo (12-factor app) in complexity - single component, boundaries clearly defined. The next steps in the real world would be - here is a payment platform with 270+ services and components, how does your AI/ML component fit into it? Who do you talk to to figure out the business value of the AI/ML component (business analysis/domain driven design)? How do you talk to your creative colleagues who are responsible for user experience in a productive manner ( i.e. jobs to be done )?

I see this gap quite consistently and I am trying to address it on the technical side by building medium size AI/ML project with 3 pipelines http://thepattern.digital/ and I think modern ML/AL professionals need to know things above before jumping into any real-world project.

Rage Against The MOOChine

I think the most important point of teaching maths for data science is to build mental models in the data scientists head. It takes time and part of the process of learning maths and usually takes 2 years (or two years course in university). Bypassing that process backfires - startups raising money for AI/ML normally take 2 years before shipping the product.

I think the mental model part is the most difficult to teach, but obviously, we are paid for specialised skills - like coding in python, hence everyone wants to jump into coding python without putting effort into learning maths and building proper mental models, which I think is wrong. The coding part is easy - I am comfortable with Octave/Matlab and Python or Rust. I would back Andrew Ng for choosing Octave for example - it's one of the most concise ways to get ML concepts working, although I disliked it when I was a student, then I tried to translate the following code into Python:

```
%% Main loop
while(gen<maxgen)
gen
% perform uniform selection, then intermediate crossover
chx=zeros(lamda,nvar); % create storage space for new chrom
sigx=zeros(lamda,nvar); % create storage space for new sigmas
for n=1:lamda % loop over all population
if(rand<xovr) % roll the dice and see if we need to cross
alpha=rand(1,nvar)*1.5-0.25; % create vector of random
% then crossover
chx(n,:)=alpha.*chrom(mod(n-1,p)+1,:)+(1-alpha).*chrom(ceil(rand*p),:);
sigx(n,:)=abs(alpha.*sig(mod(n-1,p)+1,:)+(1-alpha).*sig(ceil(rand*p),:));
else
chx(n,:)=chrom(mod(n-1,p)+1,:); % just copy if not crossing
sigx(n,:)=sig(mod(n-1,p)+1,:);
end
end
```

I build a very similar project - http://thepattern.digital/ ( full source code available https://github.com/applied-knowledge-systems/the-pattern ) and it won the platinum prize on RedisHackathon 2021.

My differentiator from connected papers is that I turn "strings into things" using external UMLC methathesaurus, so my nodes are industry-wide "concepts" and edges in the graph are a list of papers connecting the same concepts.

Shall we collaborate and take both projects further?