8 ms·
Oh I know a few. Evolution strategies, particle swarm optimization, ant colony optimization, etc, etc. They are a treasure trove of easily publishable papers,
by fivea 4y ago
Oh I know a few. Evolution strategies, particle swarm optimization, ant colony optimization, etc, etc.
They are a treasure trove of easily publishable papers, but frankly the whole field feels like a fraud. Each paper is formulaic and consists of coming up with a metaphor to add minor twists to established heuristics, and in the end they all fare slightly better than Monte Carlo.
The gravy train is provided by the No Free Lunch theorem, which is just a convenient copout to justify why an heuristics isn't good except on a single convenient realization.
- pragmatic8 4y agoThis should not have been downvoted; parent’s criticism is not unfounded. Take a look at the second section of this Wikipedia page: https://en.m.wikipedia.org/wiki/List_of_metaphor-based_metaheuristics https://en.m.wikipedia.org/wiki/List_of_metaphor-based_metah...
- matsemann 4y agoWhy not? OP asked for algorithms and explicitly said they didn't have to be state of the art. Then saying them not being state of the art is hardly relevant? And an algorithm can be based on a metaphor and still be useful, one should just take care to judge it own its own merits and not give it some greater value for being "natural".
- labarilem 4y agoYeah I know most of these algorithms are not state of the art. That's why I explicitly said that. Still, I find it interesting to read about them. It satisfies a curiosity of mine, hopefully shared with other people here too.
- lukas099 4y agoHN comments aren't StackOverflow answers. They're allowed to go off on tangents. Personally, I like to see nuances that I didn't before and so appreciated and upvoted the comment.
- mindcrime 4y agoThey are a treasure trove of easily publishable papers, but frankly the whole field feels like a fraud. Each paper is formulaic and consists of coming up with a metaphor to add minor twists to established heuristics, and in the end they all fare slightly better than Monte Carlo. My understanding is that this problem got so bad that many (most?) of the top journals in the field have declared recently(ish) that they will no longer publish EvoComp papers of that nature. That is, the ones that amount to no more than "here's a new metaphor based algorithm that doesn't really contribute anything new to our understanding of anything and represent only an incremental improvement over existing algorithms."
- ogogmad 4y agoI was under the impression that memetic algorithms for TSP (Travelling Salesman Problem) performed a lot better than expected: https://en.wikipedia.org/wiki/Memetic_algorithm https://en.wikipedia.org/wiki/Memetic_algorithm I was told this by a professor in graph theory who co-authored a book on TSP. But he had no hand in the implementation or the testing AFAICT.