Online ati Lile Negative Mining adanwo

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

1. Kilode ti ọpọlọpọ awọn odi ti a ṣe ayẹwo laileto pese ifihan agbara ikẹkọ kekere ni pipadanu mẹta?
   • Wọ́n sábà máa ń ṣi orúkọ wọn lọ́nà tí kò tọ́
   • Wọn kere ju lati baamu ni iranti
   • Wọn ti jinna tẹlẹ si oran, nitorina pipadanu ati iwọn didun wa nitosi odo
   • Wọn nigbagbogbo dogba ijinna rere
2. Kini o ṣe iyatọ iwakusa ori ayelujara lati iwakusa aisinipo?
   • Iwakusa ori ayelujara nlo intanẹẹti; offline ko
   • Iwakusa ori ayelujara yan awọn apẹẹrẹ lile laarin ipele kekere kọọkan lori fo
   • Iwakusa ori ayelujara ṣiṣẹ nikan lori ọrọ
   • Iwakusa ori ayelujara ko nilo iwe-iwọle siwaju
3. Kini odi 'ologbele-lile' bi a ti ṣalaye ni FaceNet?
   • A odi jo ju rere si oran
   • Odi ti o dọgba si oran gangan
   • Eyikeyi laileto yàn odi
   • Odi ti o jina ju rere lọ ṣugbọn sibẹ laarin ala
4. Ni iwakusa ipele-lile, odi wo ni a yan fun oran kọọkan?
   • Odi ti o sunmọ julọ ni ipele
   • Odi ti o jina julọ ninu ipele
   • A ID odi
   • Odi pẹlu ijinna agbedemeji
5. Kini idi ti iwọn ipele ti o tobi julọ ṣọ lati mu ilọsiwaju awọn abajade ikẹkọ metiriki mined?
   • O dinku iwọn ifibọ
   • O ṣe imukuro iwulo fun ala kan
   • O yọ iṣẹ isonu kuro
   • O pese adagun ti o ni oro sii ti awọn odi lile oludije fun igbesẹ kan
6. Kini eewu ti a mọ ti yiyan awọn odi ti o nira julọ?
   • Wọn ti wa ni nigbagbogbo ju rorun
   • Wọn jẹ ki matrix ipele kere
   • Wọn le jẹ aami-itọkasi tabi isunmọ awọn ẹda-ẹda ti o ba ikẹkọ jẹ
   • Wọn yọ iyọkuro kuro patapata
7. With positive distance 0.4 and margin 0.2, which negative distance is semi-hard under the described rule?
   • 0.5
   • 0.2
   • 0.8
   • 1.1
8. Why does ordinary online mining within one minibatch not necessarily find the hardest negative in the entire dataset?
   • Every batch contains the complete dataset
   • Its available candidate pool is the examples in that batch
   • Distance calculations prove no other examples exist
   • The model discards all negative labels