Amazon MLS-C01 Exam - Topic 2 Question 95 Discussion

Actual exam question for Amazon's MLS-C01 exam

Question #: 95
Topic #: 2

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?

AUpload the model to an Amazon SageMaker notebook instance and use the Amazon SageMaker HPO feature to optimize the model's hyperparameters.

BAdd more data to the training set and retrain the model using transfer learning to reduce the bias.

CUse a neural network model with more layers that are pretrained on ImageNet and apply transfer learning to increase the variance.

DTrain a new model using the current neural network architecture.

Show Suggested Answer

Suggested Answer: A

SageMaker Data Wrangler is a feature of SageMaker Studio that provides an end-to-end solution for importing, preparing, transforming, featurizing, and analyzing data. Data Wrangler includes built-in analyses that help generate visualizations and data insights in a few clicks. One of the built-in analyses is the Quick Model visualization, which can be used to quickly evaluate the data and produce importance scores for each feature. A feature importance score indicates how useful a feature is at predicting a target label. The feature importance score is between [0, 1] and a higher number indicates that the feature is more important to the whole dataset. The Quick Model visualization uses a random forest model to calculate the feature importance for each feature using the Gini importance method. This method measures the total reduction in node impurity (a measure of how well a node separates the classes) that is attributed to splitting on a particular feature. The ML developer can use the Quick Model visualization to obtain the importance scores for each feature of the dataset and use them to feature engineer the dataset. This solution requires the least development effort compared to the other options.

References:

* Analyze and Visualize

* Detect multicollinearity, target leakage, and feature correlation with Amazon SageMaker Data Wrangler

by Lavina at Jun 22, 2024, 11:17 PM

Limited Time Offer

25%

Off

Get Premium MLS-C01 Questions as Interactive Web-Based Practice Test or PDF

Contribute your Thoughts:

Submit Cancel

Phyliss

3 months ago

Not sure if just retraining with the same model will do much...

upvoted 0 times

...

Xenia

3 months ago

Definitely B, adding data should help reduce bias.

upvoted 0 times

...

Earleen

4 months ago

Wait, only 68% on training? That seems low for transfer learning.

upvoted 0 times

...

Mireya

4 months ago

I think option C could help too, more layers might capture better features.

upvoted 0 times

...

Ivette

4 months ago

More data is always a good idea!

upvoted 0 times

...

Jani

4 months ago

I’m a bit confused about the hyperparameter optimization part. Would using Amazon SageMaker really make a difference? Option A seems interesting but I’m not sure.

upvoted 0 times

...

Maryann

5 months ago

I feel like we practiced a question similar to this, and I think retraining with more data is often a solid approach. Option B sounds right to me.

upvoted 0 times

...

Xuan

5 months ago

I'm not entirely sure, but I think using a more complex model might help improve accuracy. Option C could be worth considering.

upvoted 0 times

...

Paola

5 months ago

I remember we discussed how adding more data can help reduce bias in models. So, option B seems like a good choice.

upvoted 0 times

...

Pamella

5 months ago

Alright, I've got this. The key is to increase the model's capacity without overfitting. Adding more data and optimizing the hyperparameters should do the trick. I'm feeling confident about this one.

upvoted 0 times

...

Tish

5 months ago

Okay, let's think this through. The model is already underfitting, so adding more layers or a more complex model might not be the best approach. I think focusing on the data and hyperparameters is the way to go.

upvoted 0 times

...

Jesus

5 months ago

Hmm, this is a tricky one. I'm not sure if I should try to optimize the hyperparameters or add more data. Maybe I'll try both and see what works better.

upvoted 0 times

...

Marcelle

5 months ago

I'm a bit confused here. The training and validation accuracies are pretty close, so I'm not sure if the issue is bias or variance. Maybe I should try different techniques and see what works best.

upvoted 0 times

...

Asha

5 months ago

Hmm, this looks like a networking question. I think the arp command might be the right answer since it shows the MAC addresses of devices on the local network.

upvoted 0 times

...

Albina

5 months ago

This seems pretty straightforward. I'm going to work through the math step-by-step to make sure I get the right answer. I'm confident I can solve this problem.

upvoted 0 times

...

Dortha

5 months ago

Okay, I've got this. The key is remembering that the highest-level document is the test policy, followed by the test strategy, then the project test plan, and finally the system test plan. I'm confident option A is the right answer.

upvoted 0 times

...

Kayleigh

5 months ago

I'm a bit confused here. Is a Backup Domain Controller the same as a Read-Only Domain Controller? I'll need to double-check the differences.

upvoted 0 times

...

Adelle

10 months ago

I'm leaning towards option D. Starting from scratch with a new model might be the best way to meet the 85% accuracy requirement. Can't hurt to try, right?

upvoted 0 times

Ben

9 months ago

Maybe combining both options D and B could be the best approach to try and meet the accuracy requirement.

upvoted 0 times

...

Hyman

9 months ago

I agree, more data could definitely make a difference in improving the model's performance.

upvoted 0 times

...

Malinda

9 months ago

I think option B could also be helpful. Adding more data to the training set might help reduce bias and improve accuracy.

upvoted 0 times

...

Lili

10 months ago

Ha! This is a tough one. Maybe we could try uploading the model to a SageMaker notebook and let the AI figure it out for us. Wouldn't that be a hoot?

upvoted 0 times

...

Oren

10 months ago

I disagree, I believe option C is the better choice. Using a more complex neural network model with more layers that is pretrained on ImageNet will increase the variance and potentially lead to higher accuracy.

upvoted 0 times