Laajte ci jàng bu xóot bu Bayesian

Sooy def quiz bi, tontu yi ak leeral yi dañuy feeñ.

1. Ak naka la reso neuronal bayesien di def ay poid yu wuute ak yi ñuy faral di def?
   • Dafay jàppee poid bu nekk ni distribusioŋ probabilite, du benn valeur fixe
   • Dafay jëfandikoo poid integer ci barabu desimal yi
   • Dafay dindi poid yi yépp
   • Dafay bokk benn poid ci diisaay yépp
2. What can a Bayesian approach provide that a single fixed-weight network does not directly represent?
   • A guarantee that every confidence score is calibrated
   • A distribution of predictions that reflects uncertainty about model weights
   • A proof that the model will detect every unfamiliar input
   • A way to remove all noise from the data
3. How can epistemic uncertainty change as relevant, representative training data are added?
   • It is guaranteed to reach zero with any added data
   • It is identical to label noise
   • It may decrease under the chosen model and data assumptions
   • It always increases linearly
4. Lan moo tax ñu soxla xayma yu melni inferensi variasionel wala MCMC?
   • Ndax reso neuronal yi amul benn fonction perte
   • Ndax GPU yi mënu ñu yokk
   • Ndax poid yi dañu wara nekk ay lim yu mat
   • Ndax ginaaw gi gëna jubal ci ay milioŋ ciy poid mënul nekk ci xayma
5. Lan la Monte Carlo def ngir xayma ñàkka wóor gi?
   • Dafay wéy di bàyyi dropout ci waxtu test bi, ba noppi di doxal reso bi yoon yu bari, di jëfandikoo tasaaroo ci génne yi
   • Dafay bàyyi dropout bi, di tàggat lu gëna gudd
   • Dindi bépp neuron ba fàww
   • Dafay wecci perte bi ak bruit bu bari
6. Inference variationnel dafay soppi xayma bi ci ginaaw ni ban xeetu jafe-jafe la?
   • jafe-jafe tànnal
   • Jafe-jafe gëna xéewale buy wàññi wuute KL (jaaraleko ci ELBO)
   • Jafe-jafe échantillonnage aleatoire bu amul benn mébet
   • jafe-jafe jëmmal
7. How does Bayes’ rule relate a prior, likelihood, and posterior?
   • The posterior is the prior plus the likelihood, without normalization
   • It is the likelihood divided by the prior without normalization
   • The posterior is proportional to the prior times the likelihood
   • It is the fixed point estimate of the most likely weight vector
8. How does MCMC approximate a Bayesian neural-network posterior?
   • By training a single point estimate with dropout disabled
   • By ranking model outputs with a fixed rule
   • By minimizing an ELBO with a simpler variational family
   • By drawing samples through a Markov chain whose stationary distribution targets the posterior