Idanwo ti o sopọ mọ itọsọna · Alabọde Ipele

Omo egbe Ikolu adanwo

Study training-membership inference, with shadow models, LiRA calibration, differential-privacy bounds, false positives, and attack evaluation.

Jẹmọ itọsọna awọn ọnaAwọn ikọlu Inference Ẹgbẹ
Ibeere 1 ti 8

Kini ikọlu ifọkansi ọmọ ẹgbẹ ti n gbiyanju lati pinnu?

Awotẹlẹ gbogbo ibeere8 ibeere

Awọn idahun ati awọn alaye han bi o ṣe n mu ibeere naa.

  1. Kini ikọlu ifọkansi ọmọ ẹgbẹ ti n gbiyanju lati pinnu?
    • Boya igbasilẹ kan pato jẹ apakan ti eto ikẹkọ
    • Awọn awoṣe ká išedede
    • Awọn awoṣe ká faaji
    • Awọn iwọn ti awọn igbeyewo ṣeto
  2. How can overfitting affect membership-inference exposure in studied models?
    • It proves every training record is identifiable with certainty
    • It removes the need to evaluate attack errors
    • It can increase the behavioral difference between members and comparable nonmembers
    • It makes all outputs independent of training data
  3. Kini ilana ṣe atilẹba 2017 Shokri et al. kolu gbekele?
    • Idinku gige
    • Awọn awoṣe ojiji
    • Iwadi ina
    • Àmi
  4. Kini idi ti iyasọtọ ọmọ ẹgbẹ le jẹ ipalara paapaa laisi ṣiṣafihan awọn akoonu igbasilẹ naa?
    • O fa fifalẹ awoṣe naa
    • O ayipada awọn awoṣe ká òṣuwọn
    • O npa data ikẹkọ kuro
    • Ọmọ ẹgbẹ funrararẹ le ṣe afihan awọn ododo ifura, bii ayẹwo
  5. What calibration distinguishes the cited LiRA approach from a single naive loss threshold?
    • A guarantee that every record has the same inherent difficulty
    • Only an increased generation temperature
    • Per-example score distributions estimated from in and out models
    • A requirement that the target reveal its entire training dataset
  6. What does a properly specified differential-privacy guarantee bound?
    • Changes in output probabilities between neighboring datasets
    • A guarantee that no attack can ever make any correct guess
    • A mandatory deletion of all model outputs
    • Only the raw number of files in a training folder
  7. An attack incorrectly flags 10 of 1,000 actual nonmembers as members. What is its false-positive rate on those nonmembers?
    • 10%
    • 0.1%
    • 100%
    • 1%
  8. Why does the LiRA paper emphasize true-positive rate at low false-positive rates?
    • False positives are impossible by definition
    • Average accuracy can conceal whether confidently flagged records are reliable membership evidence
    • Every member and nonmember population must have identical sizes
    • The model’s training accuracy directly gives attack precision