Idanwo ti o sopọ mọ itọsọna · Alabọde Ipele
Idanwo deede
Check regularization goals, L1 and L2 penalties, dropout, early stopping, penalty strength, and numerical examples of a regularized objective.
Ibeere 1 ti 8
Why would a practitioner add regularization when a model fits training-specific noise?
Awotẹlẹ gbogbo ibeere8 ibeere
Awọn idahun ati awọn alaye han bi o ṣe n mu ibeere naa.
- Why would a practitioner add regularization when a model fits training-specific noise?
- To guarantee zero training error
- To remove the need for validation data
- To guarantee a faster GPU kernel
- To discourage overfitting and evaluate whether unseen-data performance improves
- Ilana isọdọtun wo ni o ṣeese julọ lati wakọ diẹ ninu awọn iwuwo deede si odo, yiyan awọn ẹya ni imunadoko?
- L2 deede (ibajẹ iwuwo)
- L1 deede
- Ipese deede
- Alekun oṣuwọn ẹkọ
- Bawo ni yiyọ kuro ṣe deede nẹtiwọọki nkankikan lakoko ikẹkọ?
- O npa awọn ipele rẹ kuro patapata lati nẹtiwọki
- O ṣe afikun awọn iwọn onigun mẹrin si iṣẹ isonu
- O pa awọn neuronu laileto ni igbesẹ kọọkan ki wọn ko le ṣe deede
- O da ikẹkọ duro nigbati pipadanu afọwọsi dide
- Ninu pipadanu igbagbogbo 'aṣiṣe + lambda × ijiya', kini lambda ti o pọ si ṣe?
- Ṣe okunkun ijiya naa, titari awoṣe si awọn solusan ti o rọrun
- Irẹwẹsi ijiya ati gba awọn iwuwo nla laaye
- Yọ ọrọ aṣiṣe kuro patapata
- Ko ni ipa lori awọn iwuwo
- Kini idi ti idaduro ni kutukutu ni a le ka bii iru isọdọtun?
- Nitoripe o ṣe afikun ijiya iwuwo si isonu naa
- Nitoripe o dẹkun ikẹkọ ṣaaju ki awoṣe bẹrẹ ariwo ibamu ninu data naa
- Nitori ti o mu ki awoṣe agbara lori akoko
- Nitoripe o yọ awọn neuronu kuro laileto
- What can happen when regularization is strengthened beyond what a task needs?
- Both training and test accuracy must improve
- The penalty becomes unrelated to the objective
- Validation becomes unnecessary
- The model can underfit useful structure even if variance decreases
- For a regularization example with weights minus two and one, what are the unscaled L1 and sum-of-squares L2 penalties?
- L1 is three and L2 is five
- L1 is minus one and L2 is three
- L1 is five and L2 is three
- Both penalties are minus two
- A regularized objective has prediction error two, lambda one-half, and penalty six. What is its total value?
- Mẹjọ
- Mẹta
- Marun
- Ọkan