Feature engineering and scalability thresholds: Practice Questions — Machine Learning (NVIDIA-Certified Professional: Accelerated Data Science)

Feature Engineering and Scalability Thresholds: Practice Questions

This set of practice questions is designed to help candidates preparing for the NVIDIA-Certified Professional: Accelerated Data Science exam, focusing specifically on feature engineering and scalability thresholds within machine learning workflows accelerated by GPUs.

1. Which of the following best describes the primary goal of feature engineering in GPU-accelerated machine learning workflows?

   • A. To reduce dataset size by removing all categorical variables
   • B. To create and transform input features that improve model accuracy while maintaining computational efficiency
   • C. To increase the number of features regardless of their relevance
   • D. To avoid any preprocessing to maximize GPU utilization

   Correct answer: B

   Explanation: Feature engineering aims to create or transform features to improve model performance while considering computational efficiency, especially important when leveraging GPU acceleration.

2. When scaling a machine learning model training on GPUs, what is a common threshold that signals the need to adjust batch size or feature dimensionality?

   • A. GPU memory utilization consistently below 10%
   • B. Model accuracy plateauing despite increased training time
   • C. GPU memory exceeding capacity, causing out-of-memory errors
   • D. Using only CPU resources for feature processing

   Correct answer: C

   Explanation: Exceeding GPU memory capacity is a key scalability threshold indicating the need to optimize batch size or reduce feature dimensionality to avoid training failures.

3. Which feature engineering technique is most effective in reducing input dimensionality while preserving important information for GPU-accelerated models?

   • A. One-hot encoding of all categorical variables
   • B. Principal Component Analysis (PCA)
   • C. Adding polynomial features indiscriminately
   • D. Increasing feature sparsity without selection

   Correct answer: B

   Explanation: PCA reduces dimensionality by projecting features into a lower-dimensional space, helping manage GPU memory and computational load without losing significant information.

4. In rapid experimentation balancing accuracy and performance, what is a recommended practice related to feature engineering on GPUs?

   • A. Always use the full feature set to maximize accuracy
   • B. Start with a minimal feature set and incrementally add features while monitoring GPU resource usage and model accuracy
   • C. Ignore GPU memory constraints during feature selection
   • D. Use random feature subsets without validation

   Correct answer: B

   Explanation: Incrementally adding features allows balancing accuracy gains against GPU resource usage, enabling efficient rapid experimentation.

5. Which scenario indicates that feature engineering has reached a scalability threshold in a multi-GPU training environment?

   • A. Training time decreases linearly with added GPUs
   • B. Communication overhead between GPUs increases disproportionately due to large feature sets
   • C. Model accuracy improves with more features
   • D. GPU memory usage remains constant

   Correct answer: B

   Explanation: Excessive feature dimensionality can increase inter-GPU communication overhead, limiting scalability and efficiency in multi-GPU setups.

6. Which GPU memory technique helps manage large feature sets during training without sacrificing model accuracy?

   • A. Using mixed precision to reduce memory footprint
   • B. Disabling batching to process one sample at a time
   • C. Converting all features to 64-bit floats
   • D. Avoiding any feature normalization

   Correct answer: A

   Explanation: Mixed precision uses lower-precision data types where possible, reducing GPU memory usage and enabling larger feature sets or batch sizes without losing accuracy.

7. What is a key consideration when performing feature engineering to ensure scalability in GPU-accelerated machine learning?

   • A. Maximizing the number of features regardless of their relevance
   • B. Ensuring features are engineered to minimize memory usage and computational complexity
   • C. Avoiding any feature transformation to reduce preprocessing time
   • D. Using only CPU-based feature engineering methods

   Correct answer: B

   Explanation: Scalability requires features to be engineered efficiently to fit within GPU memory and computational limits, balancing model complexity and resource constraints.

More in this topic

Feature engineering and scalability thresholds: Common Mistakes — Machine Learning (NVIDIA-Certified Professional: Accelerated Data Science)Single- and multi-GPU training: Common Mistakes — Machine Learning (NVIDIA-Certified Professional: Accelerated Data Science)Single- and multi-GPU training — Machine Learning (NVIDIA-Certified Professional: Accelerated Data Science)GPU memory techniques such as batching and mixed precision: Quick Reference — Machine Learning (NVIDIA-Certified Professional: Accelerated Data Science)Single- and multi-GPU training: Quick Reference — Machine Learning (NVIDIA-Certified Professional: Accelerated Data Science)GPU memory techniques such as batching and mixed precision: Worked Example — Machine Learning (NVIDIA-Certified Professional: Accelerated Data Science)Rapid experimentation balancing accuracy and performance — Machine Learning (NVIDIA-Certified Professional: Accelerated Data Science)Feature engineering and scalability thresholds: Quick Reference — Machine Learning (NVIDIA-Certified Professional: Accelerated Data Science)Machine Learning — NVIDIA-Certified Professional: Accelerated Data ScienceGPU memory techniques such as batching and mixed precision — Machine Learning (NVIDIA-Certified Professional: Accelerated Data Science)Hyperparameter optimization — Machine Learning (NVIDIA-Certified Professional: Accelerated Data Science)Feature engineering and scalability thresholds: Worked Example — Machine Learning (NVIDIA-Certified Professional: Accelerated Data Science)Single- and multi-GPU training: Practice Questions — Machine Learning (NVIDIA-Certified Professional: Accelerated Data Science)GPU memory techniques such as batching and mixed precision: Practice Questions — Machine Learning (NVIDIA-Certified Professional: Accelerated Data Science)GPU memory techniques such as batching and mixed precision: Common Mistakes — Machine Learning (NVIDIA-Certified Professional: Accelerated Data Science)Feature engineering and scalability thresholds — Machine Learning (NVIDIA-Certified Professional: Accelerated Data Science)Single- and multi-GPU training: Worked Example — Machine Learning (NVIDIA-Certified Professional: Accelerated Data Science)