Bird
0
0

Why does GloVe use a weighted least squares objective instead of just factorizing the co-occurrence matrix directly?

hard📝 Conceptual Q10 of 15
NLP - Word Embeddings
Why does GloVe use a weighted least squares objective instead of just factorizing the co-occurrence matrix directly?
ABecause direct factorization is computationally impossible
BTo reduce the impact of very frequent word pairs and emphasize informative co-occurrences
CTo ensure embeddings are sparse vectors
DTo avoid using any matrix operations
Step-by-Step Solution
Solution:

  1. Step 1: Understand GloVe's weighting function

    GloVe uses weights to limit influence of very frequent pairs that add noise.

  2. Step 2: Reason about objective choice

    Weighted least squares emphasize informative co-occurrences and reduce bias from common words.

  3. Final Answer:

    To reduce the impact of very frequent word pairs and emphasize informative co-occurrences -> Option B

  4. Quick Check:

    Weighting balances frequent vs informative pairs [OK]
Quick Trick: Weighting reduces noise from frequent pairs [OK]
Common Mistakes:
MISTAKES
  • Thinking direct factorization is impossible
  • Assuming embeddings must be sparse
  • Believing matrix ops are avoided

Want More Practice?

15+ quiz questions · All difficulty levels · Free

Free Signup - Practice All Questions
More NLP Quizzes
Sentiment Analysis Advanced - Hybrid approaches - Quiz 1easy Text Generation - Beam search decoding - Quiz 2easy Text Generation - N-gram language models - Quiz 9hard Text Similarity and Search - Information retrieval basics - Quiz 15hard Topic Modeling - Visualizing topics (pyLDAvis) - Quiz 7medium Topic Modeling - Topic coherence evaluation - Quiz 1easy Topic Modeling - LDA with scikit-learn - Quiz 2easy Topic Modeling - LDA with scikit-learn - Quiz 13medium Word Embeddings - Pre-trained embedding usage - Quiz 2easy Word Embeddings - Word2Vec (CBOW and Skip-gram) - Quiz 5medium