AML Natural Language Processing & Computer Vision Flashcards

Which NLP preprocessing step reduces words to their base or root form by removing suffixes (e.g., 'running' → 'run')?

Answer: Stemming

Stemming applies rule-based suffix stripping to reduce words to an approximate root, which may not be a valid dictionary word.

What is semantic segmentation in computer vision?

Answer: Assigning a class label to every pixel in an image

Semantic segmentation classifies each pixel of an image into a category, producing a pixel-wise class map of the entire scene.

In NLP, what does Named Entity Recognition (NER) identify in text?

Answer: Real-world entities such as people, organizations, and locations

NER tags spans of text that refer to specific entities like person names, companies, dates, and geographic locations.

Which architecture is most commonly used for image-to-image tasks like semantic segmentation and medical image analysis?

Answer: U-Net (encoder-decoder with skip connections)

U-Net uses a contracting encoder path and an expanding decoder path with skip connections that preserve spatial detail for precise pixel prediction.

What is the purpose of subword tokenization methods like Byte-Pair Encoding (BPE) used in models like GPT?

Answer: Balance vocabulary size and out-of-vocabulary handling by splitting rare words into frequent subword units

BPE iteratively merges frequent character pairs into subword units, allowing models to handle rare and unseen words by decomposing them.

Which evaluation metric for generative language models measures how well a probability model predicts a sample by computing the exponential of average negative log-likelihood?

Answer: Perplexity

Perplexity quantifies how surprised a language model is by new text; lower perplexity indicates better predictive performance.