Would maps need the same or different presets than Image Occlusions?

As I was making my umpteenth Image Occlusion (IO from now on) card I started to wonder if they should have the same FSRS preset as the cards containing (geography) maps because (my reasoning is) whether I make an IO about motherboard parts, IO about anatomy, or Basic (front-back) card about Turkish regions and provinces, they all intuitively seem to go through the same mental pathways, as it’s about recognizing part of an image and recalling its name.

I’m not really sure about this reasoning which is why I’d like to have your input on this. Would you clump them all in the same preset or have different presets for maps and IO and what’s your reasoning? Also, do you know of any resource or person talking about this kind of things?

1 Like

Interesting question. My intuition makes me agree with you. Following.

I have no known good method to determine how to divide your cards but generally you should focus on how difficult something feels to you.

FSRS is still very new and I haven’t seen anyone talking about this (except us here, but we of course don’t write blogs and articles).

FSRS’s tutorial has a relevant section, written by Expertium, mostly from my inputs: [I’ve included an easier method down below]

If you are confident in your knowledge of the algorithm and are willing to do extra work, there is a way to check whether a deck benefits from having it’s own preset.

  • Go to deck options of the deck you want to create a new preset for.
  • Now copy the parameters from the FSRS parameters field. We will use this later for evaluation.
  • Now create a new preset and save it to this deck.
  • Paste the parameters you previously copied in the FSRS parameters field of the newly created preset.
  • Click Evaluate and write down the RMSE and log loss values.
  • Now click Optimize to obtain a new set of parameters. Click Evaluate and write down the RMSE and log loss values.
  • If this new set of parameters results in lower RMSE and log loss values, then save the preset along with the new parameters.

An easier way to do this is (@Expertium do you want to have this in the tutorial?) —

  1. Go to the deck you want to have a new preset for.
  2. Click on clone preset and save.
  3. Now click optimize.

Note: If you have previously inserted any text in the search bar just above the optimize button, you might have to clear that up to run optimisation.

If optimisation works you have improved your RMSE and log loss values. Spoiler: If you have enough number of revlogs it will almost always work. I just some hours ago created a lot of new presets for my Kanken (an exam) deck. I saw huge improvements in my Evaluation values. Also something a bit more concrete was, my huge backlog was reduced (In one subdeck, from 85 cards to 30) as I was using reschedule cards on change.

Note that all these cards test me on the same type of material (writing Kanji characters). One thing I noticed was my lower level decks’ initial stabilities got bumped up a lot than before. So I might have been overdoing those easier cards just because the previous preset covered a lot of harder cards.

https://www.reddit.com/r/Anki/comments/18jvyun/some_posts_and_articles_about_fsrs/

1 Like

My bad, I thought you wanted to learn more about the algorithm.

Well, I don’t have resources specifically about maps, but I do have these:

  1. https://controlaltbackspace.org
  2. https://andymatuschak.org
  3. https://augmentingcognition.com
  4. https://eshapard.github.io (inactive since 2022)
  5. https://cademcniven.com
  6. https://supermemo.guru (a lot of random ramblings)
  7. https://gwern.net/spaced-repetition
    (reluctantly) https://www.astralcodexten.com and https://slatestarcodex.com/
    (VERY reluctantly) https://www.lesswrong.com/

These are blogs by people who mentioned Anki and/or spaced repetition at least once and wrote about other intellectual stuff. The last one, lesswrong.com, is more like a collection of blogs by different people rather than a blog made by a single person + like half of the posts are about AI + they have a metric crapton of jargon developed over the years + let’s just say, lesswrong.com attracts a specific type of person, the kind of person who, if you called him a “nerd”, would write a 2000-word essay and cite 20 papers in his response to you.

2 Likes

If you mean locator maps, so passively recalling the name of X part highlighted in a map, that is passive review rather than active recall. I have done the latter, a blank map with the boarders and locate X as the front side.

I used this one for Spain File:Spain location map with provinces.svg - Wikipedia

That predates FSRS for more than a decade though. My anecdotical evidence is quite obvious: passive recognition cards tend to be easier and can have longer intervals

1 Like