controller POMDP policies with FST. Previous papers had exponential blowups.

Successor function is deterministic.

policy iteration

Use FST as policy representation:

1. deterministic controller POMDP evaluation
2. for all \((a,o,x)\), add a now node x’ and evaluate them to see if its needed
3. then, we perform pruning
   • everything that’s dominated (i.e. \(U(x,s) < U(x’, s) \forall s\). i.e. we want to prune everything for which the expected utility of being in node \(x’\) dominates the expected utility of \(x\) for all \(x\).
   • prune new nodes that are duplicates in terms of action and transitions

When you are done, extract the policy: find the node that maximizes your

Haplmmune is a antibody platform technology developed by Akiko Koide (NYU) specific towards ?

Theorem: P != NP IFF P = BPP Theorem’: if SAT requires exponential time, then, we can show that P = BPP.

Harmonic Mean is the inverse of the inverse sum of arithmetic means, weighted.

It is near the lower of the two values, instead of the middle: meaning that in incentives both things being meaned to be higher. Hence why we use F measure for things.

Representational Harms

System’s representation demeans a social group because it learns about built-in biaes of data

Harms of Censorship

Speech that mention minority group gets sensored because they mention minority groups.

Performance Disparities

For instance, works worse on AAVE. Lack of data, labels, etc.