Bending is what happens when you apply a transverse load to an object and it goes wooosh.

That’s cool. Now how does it work? see Euler-Bernoulli Theory

Consider a case where there’s only a single binary outcome:

• “success”, with probability \(p\)
• “failure”, with probability \(1-p\)

constituents

\begin{equation} X \sim Bern(p) \end{equation}

requirements

the probability mass function:

\begin{equation} P(X=k) = \begin{cases} p,\ if\ k=1\\ 1-p,\ if\ k=0\\ \end{cases} \end{equation}

This is sadly not Differentiable, which is sad for Maximum Likelihood Parameter Learning. Therefore, we write:

\begin{equation} P(X=k) = p^{k} (1-p)^{1-k} \end{equation}

Which emulates the behavior of your function at \(0\) and \(1\) and we kinda don’t care any other place.

\begin{equation} x^{2}y’’ + xy’ + (x^{2}-n^{2})y = 0 \end{equation}

this function is very useful, they have no well defined elementary result.

best-action worst-state is a lower bound for alpha vectors:

\begin{equation} r_{baws} = \max_{a} \sum_{k=1}^{\infty} \gamma^{k-1} \min_{s}R(s,a) \end{equation}

The alpha vector corresponding to this system would be the same \(r_{baws}\) at each slot.

which should give us the highest possible reward possible given we always pick the most optimal actions while being stuck in the worst state

Background

recall AlphaZero

1. Selection (UCB 1, or DTW, etc.)
2. Expansion (generate possible belief notes)
3. Simulation (if its a brand new node, Rollout, etc.)
4. Backpropegation (backpropegate your values up)

Key Idea

Remove the need for heuristics for MCTS—removing inductive bias

Approach

We keep the ol’ neural network:

\begin{equation} f_{\theta}(b_{t}) = (p_{t}, v_{t}) \end{equation}

Policy Evaluation

Do \(n\) episodes of MCTS, then use cross entropy to improve \(f\)

Ground truth policy

Action Selection

Uses Double Progressive Widening