NCP-ADS 試験問題を無料オンラインアクセス

You are performing data cleansing on a large dataset using CuDF. The dataset contains numerical values, some of which are outliers. You need to remove or adjust these outliers to make your model training more robust.
Which of the following approaches should you consider for handling outliers efficiently in CuDF? (Select two)

• A.Using clip() to set a maximum and minimum threshold for numerical values
• B.Using quantile() to calculate the interquartile range (IQR) and filter out outliers
• C.Using dropna() to remove rows with outliers
• D.Using applymap() to apply a custom function for handling outliers

Which of the following statements best describes the role of GPUs in accelerating data science workloads?

• A.GPUs are only effective for acceleration when used in conjunction with Tensor Processing Units (TPUs), as they cannot train deep learning models independently.
• B.GPUs are optimized for sequential data processing tasks, making them more efficient than CPUs for database operations.
• C.GPUs use thousands of smaller cores that can execute many parallel computations simultaneously, making them ideal for large-scale matrix operations.
• D.GPUs are designed primarily for rendering graphics and have limited utility in machine learning and deep learning applications.

You are working on an MLOps pipeline that involves loading a large dataset for training a deep learning model on an NVIDIA GPU. Before training, you need to ensure that the dataset fits within the available GPU memory.
Which of the following commands in Python using the pandas and numpy libraries can correctly determine the memory size of a dataset?

• A.df.memory_usage(deep=True).sum()
• B.df.info(memory_usage='deep')
• C.sys.getsizeof(df)
• D.np.array(df).nbytes

You are training a machine learning model using NVIDIA RAPIDS cuML and notice that the training process is significantly slower than expected. You suspect that there are bottlenecks in data movement and computation.
Which of the following techniques can best help you diagnose and resolve these bottlenecks?

• A.Reduce the number of features used in training without profiling the actual bottlenecks.
• B.Use cudf.DataFrame.to_pandas() to convert the dataset to a pandas DataFrame for analysis.
• C.Use cuml.common.device_auto_mem_size() to check GPU memory usage and adjust batch sizes accordingly.
• D.Move all data from GPU memory to CPU memory before training the model.

You are processing a dataset of high-resolution images for deep learning training and want to optimize image loading, augmentation, and transformation on an NVIDIA GPU.
Which of the following statements correctly describes the role of NVIDIA DALI (Data Loading Library) in accelerating data preparation?

• A.NVIDIA DALI can offload image preprocessing tasks to the GPU, reducing CPU bottlenecks and improving training throughput.
• B.NVIDIA DALI only supports preprocessing images and does not include any functionality for video or text data.
• C.NVIDIA DALI is an alternative to cuDF for GPU-accelerated tabular data processing.
• D.NVIDIA DALI does not support data augmentation and is only used for raw data loading.