MLA-C01 Question 416
Single answerYou are building a machine learning model to perform real-time image classification on a large dataset of high-resolution images. The model requires low latency inference and benefits from hardware acceleration. Which instance type is most suitable for deploying your model in this scenario?
- A
Compute optimized instances (e.g., C5)
- B
Memory optimized instances (e.g., R5)
- C
Inference optimized instances (e.g., Inf1)
- D
General purpose instances (e.g., M5)
Show answer and explanation
Correct answer: C
Explanation
Inference optimized instances (e.g., Inf1) are designed for machine learning inference use cases, providing low latency and high throughput using AWS Inferentia chips. This makes them ideal for real-time image classification tasks where both latency and performance are critical. Other instance types, such as compute optimized or memory optimized, are better suited for workloads with specific CPU or memory demands but lack the specialized hardware for efficient inference.
- A. Incorrect.
Compute optimized instances (e.g., C5) are best suited for workloads requiring high CPU performance, such as batch processing or computational simulations. They are not optimized for low-latency inference or hardware acceleration.
- B. Incorrect.
Memory optimized instances (e.g., R5) are ideal for workloads requiring high memory capacity, such as large in-memory databases or big data analytics. They are not designed for low-latency inference or hardware acceleration.
- C. Correct.
Inference optimized instances (e.g., Inf1) are specifically designed for machine learning inference workloads. They provide low latency and hardware acceleration through AWS Inferentia chips, making them the best choice for real-time image classification tasks.
- D. Incorrect.
General purpose instances (e.g., M5) offer a balance of compute, memory, and networking resources, suitable for a broad range of applications. However, they lack the specialized hardware and optimization for low-latency inference.