3 ms·
Adversary input is basically at the core of analysis of these types of algorithms, yes. Here's another crossover analogy. In games, for any deterministic strat
by jvanderbot 1y ago
Adversary input is basically at the core of analysis of these types of algorithms, yes.
Here's another crossover analogy. In games, for any deterministic strategy there are games where you can't play a deterministic (pure) strategy and hope to get the best outcome, you need randomized (mixed) ones. Otherwise the Adversary could have anticipated and generated a different pathological response.
- hinkley 1y agoI think this started though with systems doing incremental work and then passing the result on to an algorithm that does worse with partially sorted or inverted lists than with random ones. Adversarial came later, particularly with the spread of the Internet.
- jvanderbot 1y agoI assure you adversarial analysis predates the internet