MLS-C01 Question 145
Single answerYou are tasked with building a machine learning model to classify customer complaints into multiple categories for a support ticketing system. The dataset is textual and consists of thousands of labeled complaints. You need a solution that balances accuracy and training efficiency while leveraging pre-trained knowledge from large datasets. Which approach is the most suitable for this scenario?
- A
Train a Random Forest classifier on the raw text data.
- B
Use k-means clustering to group similar complaints.
- C
Fine-tune a pre-trained Large Language Model (LLM) for text classification.
- D
Train a Convolutional Neural Network (CNN) from scratch on the text data.
Show answer and explanation
Correct answer: C
Explanation
Fine-tuning a pre-trained Large Language Model (LLM) is the most appropriate approach for this scenario because it allows you to leverage transfer learning. LLMs, such as BERT or GPT, have been trained on vast amounts of textual data, making them highly effective for NLP tasks. Fine-tuning these models on your specific dataset provides a balance of accuracy and efficiency, especially for text classification tasks with labeled data.
- A. Incorrect.
Random Forest classifiers are not well-suited for raw textual data unless the text is explicitly transformed into numerical features (e.g., through TF-IDF or embeddings). Even then, they may not achieve state-of-the-art accuracy for complex NLP tasks.
- B. Incorrect.
k-means clustering is an unsupervised learning algorithm and is useful for grouping similar data points. However, it is not suitable for supervised classification tasks like this one.
- C. Correct.
Fine-tuning a pre-trained Large Language Model (LLM) leverages transfer learning, allowing the model to use its prior knowledge from large textual datasets. This approach provides high accuracy and efficiency for text classification tasks.
- D. Incorrect.
Training a CNN from scratch for text classification requires a large amount of labeled data and computational resources. It is less efficient and less accurate than fine-tuning pre-trained models for this scenario.