Iyatọ-Iyatọ Tradeoff adanwo

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

1. Awoṣe ti o rọrun pupọ lati mu ilana ti o wa labẹ data ni akọkọ jiya lati:
   • Iyatọ giga
   • Iyatọ giga
   • Aṣiṣe irreducible kekere
   • Aṣeju pupọ
2. Oju iṣẹlẹ wo ni ami Ayebaye ti iyatọ giga (overfitting)?
   • Ikẹkọ ati aṣiṣe idanwo jẹ mejeeji giga ati iru
   • Awọn awoṣe foju julọ awọn ẹya ara ẹrọ
   • Ṣafikun isọdọtun diẹ sii mu aṣiṣe ikẹkọ pọ si si aṣiṣe idanwo
   • Aṣiṣe ikẹkọ jẹ kekere pupọ ṣugbọn aṣiṣe idanwo jẹ ga julọ
3. In the classic bias-variance picture described in the guide, which tradeoff can accompany increased model flexibility?
   • Iyatọ n dinku ati iyatọ pọ si
   • Awọn mejeeji pọ si
   • Mejeeji dinku
   • Iyatọ n pọ si ati iyatọ dinku
4. Ṣafikun isọdọtun L2 (oke) si awoṣe ni gbogbogbo:
   • Dinku ojuṣaaju ni idiyele ti iyatọ diẹ sii
   • Imukuro aṣiṣe ti ko le dinku
   • Dinku iyatọ ni idiyele ti irẹjẹ diẹ diẹ sii
   • Ko ni ipa lori boya irẹjẹ tabi iyatọ
5. Aṣiṣe aibikita ninu jijẹ aiṣedeede-iyatọ duro:
   • Aṣiṣe ṣẹlẹ nipasẹ yiyan algorithm ti ko tọ
   • Aṣiṣe lati ariwo atorunwa ninu data ti ko si awoṣe le yọkuro
   • Aṣiṣe ti o padanu pẹlu data ikẹkọ diẹ sii
   • Aṣiṣe ti o ṣẹlẹ nipasẹ iyatọ giga nikan
6. Ìṣẹ̀lẹ̀ ‘ìsọkalẹ̀ ìlọ́po méjì’ tí a ṣàkíyèsí nínú ẹ̀kọ́ ìjìnlẹ̀ dábàá pé:
   • Aṣiṣe idanwo le dinku lẹẹkansi bi awọn awoṣe ṣe dagba ti o ti kọja aaye ti interpolating data ikẹkọ
   • Aṣiṣe idanwo nigbagbogbo tẹle apẹrẹ U ti o rọrun
   • Iyatọ ko ṣe pataki fun awọn nẹtiwọọki nkankikan
   • Irẹjẹ nigbagbogbo jẹ odo ni awọn awoṣe ti o jinlẹ
7. In a squared-error decomposition, squared bias is 2, variance is 3, and irreducible noise is 1. What is the total expected error?
   • 14
   • 5
   • 3
   • 6
8. Why can averaging many de-correlated trees help in the random-forest example?
   • It removes all noise from the data-generating process
   • It guarantees zero test error
   • It can reduce the variance of the combined prediction
   • It converts every tree into a linear model