Exam AWS Certified Machine Learning - Specialty All Questions

View all questions & answers for the AWS Certified Machine Learning - Specialty exam

Exam AWS Certified Machine Learning - Specialty topic 1 question 98 discussion

Exam question from Amazon's AWS Certified Machine Learning - Specialty

Question #: 98
Topic #: 1

[All AWS Certified Machine Learning - Specialty Questions]

A machine learning specialist works for a fruit processing company and needs to build a system that categorizes apples into three types. The specialist has collected a dataset that contains 150 images for each type of apple and applied transfer learning on a neural network that was pretrained on ImageNet with this dataset.
The company requires at least 85% accuracy to make use of the model.
After an exhaustive grid search, the optimal hyperparameters produced the following:
✑ 68% accuracy on the training set
✑ 67% accuracy on the validation set
What can the machine learning specialist do to improve the system's accuracy?

A. Upload the model to an Amazon SageMaker notebook instance and use the Amazon SageMaker HPO feature to optimize the model's hyperparameters.
B. Add more data to the training set and retrain the model using transfer learning to reduce the bias.
C. Use a neural network model with more layers that are pretrained on ImageNet and apply transfer learning to increase the variance.
D. Train a new model using the current neural network architecture.

Show Suggested Answer

Suggested Answer: B 🗳️

by Istdanagan at April 21, 2022, 5:09 a.m.

Disclaimers:

- ExamTopics website is not related to, affiliated with, endorsed or authorized by Amazon.
- Trademarks, certification & product names are used for reference only and belong to Amazon.

Comments

Submit Cancel

dolorez

Highly Voted 2 years, 1 month ago

Selected Answer: B

the answer is B - the model is underfitting = high bias, so we want to reduce it C is wrong because the intention is not to increase variance which equals overfitting (using a more complex model would be good, but to reduce bias not increase variance)

upvoted 11 times

...

CloudGyan

Most Recent 6 months ago

Selected Answer: B

The 68% accuracy on the training set and 67% accuracy on the validation set suggest that the model is biased - underfitting and does not have enough capacity or relevant information to learn the underlying patterns in the data.

upvoted 1 times

...

endeesa

7 months, 2 weeks ago

Selected Answer: B

I would think ImageNet network is good enough already, so more data

upvoted 1 times

...

loict

10 months ago

Selected Answer: B

A. NO - HPO has already been done though grid search B. YES - 150 images is very small; need x10 that C. NO - need bigger training set D. NO - what would the new model be ?

upvoted 2 times

...

Mickey321

10 months, 2 weeks ago

Selected Answer: B

More data to training set

upvoted 1 times

...

kaike_reis

11 months, 2 weeks ago

Selected Answer: B

Letter B is the correct one. We can add more data with data augmentation. Letter A would be a repetition of what has already been done. Letter C is impractical. Letter D is starting from scratch without need.

upvoted 1 times

...

mirik

1 year ago

Selected Answer: D

I think it should be D: "Train a new model using the current neural network architecture". Because apples data is very specific and ImageNet weights will be to generic there. We still can leave ImageNet weights for an initial configuration but the model should be retrained from scratch.

upvoted 1 times

...

cox1960

1 year, 2 months ago

Selected Answer: A

450 images should be fine. HPO for me.

upvoted 1 times

...

expertguru

1 year, 6 months ago

bOTH VALIdation set and train set performing equally but performance not good. So the basic problem here is high bias (train error) and high variance (test error). Ideally we want both low, but there is trade-off need to be cautious to avoid overfitting. So this problem needs solution for Low bias first (so training performance improves with decent) for later to figure out whether that leads to overfit or not when you test it,! Answer choice B

upvoted 1 times

...

deng113jie

1 year, 11 months ago

why not A? https://aws.amazon.com/about-aws/whats-new/2022/07/amazon-sagemaker-automatic-model-tuning-supports-increased-limits-improve-accuracy-models/

upvoted 2 times

...

edvardo

2 years, 1 month ago

Given that the model can't even fit the training set properly, it would be convenient to amplify the layers that are trained. If I understood the phrasing correctly, I would go with C.

upvoted 1 times

...