Sparsifying Knowledge-Graph using Target Information

Sparsifying Knowledge-Graph Using Target Information

Sparsifying knowledge graphs for supervised tasks, using PMI to remove irrelevant edges; with concrete example using medical data

“Medical Knowledge Graph” — generated at deepai.org

The following is a simplified snippet from a research project I conducted while working at Microsoft Research, under the guidance of Professor Elad Yom-Tov.

Prerequisites: Familiarity with supervised machine learning and a basic grasp of Knowledge Graphs, including their utilization in Feature Engineering.

TL;DR

You have a Supervised ML …

concrete data example feature engineering generated graph graphs guidance information knowledge knowledge graph knowledge-graph-embedding knowledge graphs machine medical medical data microsoft microsoft research professor project research simplified supervised learning tasks while