Exam AWS Certified Machine Learning - Specialty topic 1 question 145 discussion

A should be the answer

BlazingText can also do supervised text classification

Built-in BlazingText model using Word2Vec mode in Amazon SageMaker would likely be quicker to set up compared to using Amazon Comprehend for this specific use case. Since the problem statement mentions that the review data is already labeled with the correct durability result, preparing the training data should be relatively straightforward. Additionally, as a built-in algorithm, BlazingText is optimized and pre-configured for text classification tasks, reducing the need for extensive customization and configuration compared to using Amazon Comprehend for this specific use case. It's important to note that while BlazingText may be quicker to set up for this particular task, Amazon Comprehend offers a broader range of NLP capabilities and may be more suitable for other NLP tasks or scenarios where more customization and flexibility are required. However, given the time constraint of 2 days and the specific requirement of identifying product durability concerns from reviews, training a built-in BlazingText model using Word2Vec mode in Amazon SageMaker is likely to be the more direct and quicker approach to get a working solution set up and running.

Given the time constraint of 2 days and the need for a quick solution, the most direct approach would be to choose an option that provides a ready-to-use solution without the need for extensive customization or training. Among the given options, the most direct approach would be: C. Train a built-in BlazingText model using Word2Vec mode in Amazon SageMaker. This option allows you to leverage a pre-built model (BlazingText) that is optimized for text classification tasks. Word2Vec mode is suitable for analyzing text data and can quickly provide insights into sentiment or, in this case, concerns over product durability. This approach minimizes the need for extensive data preprocessing and model tuning, allowing you to focus on training and deploying the model within the given timeframe.

Using a existing model to do the task in 2 days. A

I would say A

A. YES - Amazon Comprehend with multi-class mode and Augmented manifest file B. NO - Gluon is for timeseries C. NO - still a lot of work after generating embedding D. NO - seq2seq is to generate text, we want to classify

To solve the problem in 2 days, and dealing with sentiment analysis so A will be the right answer using the comprehend AWS Comprehend is a natural language processing (NLP) service that uses machine learning to discover insights from text. It provides a range of functionalities, including detecting language and sentiment, extracting named entities and key phrases, and tagging parts of speech5. AWS Comprehend can automatically break down concepts like entities, phrases, and syntax in a document, which is particularly helpful for identifying events, organizations, persons, or products referenced in a document

The most direct approach to solve this problem within 2 days is option A, train a custom classifier by using Amazon Comprehend. By doing so, you can use Amazon Comprehend, a natural language processing (NLP) service that uses machine learning to find insights and relationships in text, to create a custom classifier that can identify reviews expressing concerns over product durability. You can use the labeled reviews as your training data and specify the durability result as the class label. Amazon Comprehend will automatically preprocess the text, extract features, and train the classifier for you. You can also use Amazon Comprehend to evaluate the performance of your classifier and deploy it as an endpoint. This way, you can train a model to solve this problem within 2 days without requiring much coding or infrastructure management.

A: You can customize Amazon Comprehend for your specific requirements without the skillset required to build machine learning-based NLP solutions. Using automatic machine learning, or AutoML, Comprehend Custom builds customized NLP models on your behalf, using training data that you provide.

Comprehend can do Custom Classification

Comprehend can do Sentiment Analysis

The answer is C, because of the amount of data, and the time constraint. C is the most efficient solution. Conventionally A would be the right answer, but given the time constraint the answer is C.

I would say blaze text. Cuz comprehend needs custom code, so we have only 2 days.

https://docs.aws.amazon.com/comprehend/latest/dg/how-document-classification.html

If the problem needs to be solved in 2 days I would avoid going with any customised solution which would eliminate A and B. As the data is labelled already we don't need an unsupervised algorithm therefore eliminating C. Which leaves us with D