CS 7643 QUIZ 4 ACTUAL 40 QUESTIONS AND

CS 7643 QUIZ 4 (ACTUAL / ) 40 QUESTIONS AND

VERIFIED ANSWERS/UPDATE| 100%

Evaluating Word Embeddings Extrinsic - --aNSWERS---- Evaluation on real task

• Can take a long time to compute
• Unclear if the subsystem is the problem or its interaction
• if replacing exactly one subsystem with another improves

accuracy -> winning

Why Graph Embeddings - --aNSWERS---- They are a form of unsupervised learning on graphs

• Results in task-agnostic entity representations
• Features are useful on downstream tasks without much data
• Nearest Neighbors are semantically meaningful

Graph Embeddings Loss Function - --aNSWERS---- Margin loss between the score of an edge f(e) and a negative sampled edge f(e')

• Negative sampled edges are constructed by taking real edge

and replacing either the source or destination vertex with a random node

• the score of an edge f(e) is a similarity (dot product) between

the source embedding and a transformed version of the destination embedding 1 / 2

• f(e) = cos( theta(s) , theta(d) + theta(r) )

Graph Embedding is Slow: Reason and Solution - --aNSWERS-

--- Training time dominated by computing scores for "fake edges"

• Corrupt a sub-batch of edges with the same set of random

nodes

Debiasing word2vec - --aNSWERS---- identify gender subspace with gendered words

• project all words onto this subspace
• subtract those projections from the original word

Problem: Not that effective and bias pervades the word

embedding space

t-SNE things to remember - --aNSWERS---1. Run until it stabilizes

• Set perplexity b/w 2 and N
• perplexity loosely measures # neighbors
• balances b/w local and global aspects of nodes

• Re-run t-SNE multiple times to ensure we get the same

shape

• / 2