Awọn ilana Gaussian adanwo

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

1. Ilana Gaussian kan n ṣalaye pinpin iṣeeṣe lori kini?
   • Awọn iṣẹ
   • Awọn aaye data kọọkan nikan
   • Oye kilasi akole
   • Awọn iwuwo nẹtiwọọki nkankikan
2. Beyond a point estimate, what does a Gaussian Process return under its chosen kernel and noise assumptions?
   • Only an exact parameter vector
   • A posterior predictive distribution with an uncertainty estimate
   • A guaranteed calibrated interval under any model mismatch
   • A deterministic label with no variance
3. Ipa wo ni ekuro (iṣẹ iṣọpọ) ṣe ninu GP kan?
   • O ṣe aami data ikẹkọ
   • O yọ iwulo fun data ikẹkọ kuro
   • O ṣe koodu bi awọn abajade ti o jọra ṣe jẹ fun awọn igbewọle nitosi
   • O ṣeto oṣuwọn ẹkọ
4. Kini idi ti Awọn ilana Gaussian boṣewa n tiraka pẹlu awọn ipilẹ data ti o tobi pupọ?
   • Wọn ko le ṣe aṣoju awọn iṣẹ aiṣedeede
   • Wọn nilo awọn kilasi aami
   • Wọn foju kọ data ikẹkọ patapata
   • Iyipada matrix ekuro ni idiyele nipa akoko n-cubed
5. Bawo ni awọn hyperparameters GP gẹgẹbi iwọn gigun ti a maa n yan?
   • Nipa mimu ki o ṣeeṣe ti o pọju
   • Nipa lafaimo lairotẹlẹ
   • Nipa dindinku awọn nọmba ti data ojuami
   • Wọn ti wa ni titunse nigbagbogbo ati ki o ko aifwy
6. Ninu ilana ti o gbajumọ ni Awọn ilana Gaussian jẹ paati mojuto?
   • Convolutional image classification
   • Bayesian iṣapeye fun hyperparameter tuning
   • Tokenization ti ọrọ
   • Backpropagation nipasẹ jin nẹtiwọki
7. Which kernel is described in the guide as smoother than the Matérn alternative?
   • A unigram frequency kernel
   • A transition-probability kernel
   • The RBF (squared exponential) kernel
   • The Viterbi kernel
8. Why can a GP’s predictive uncertainty widen far from training observations?
   • The model adds more observed labels automatically
   • The mean function becomes a neural network
   • The kernel matrix becomes smaller than zero
   • There is less nearby data constraining the function under the selected covariance assumptions