Stanford CS224W: ML with Graphs | 2021 | Lecture 10.3 - Knowledge Graph Completion Algorithms

preview_player
Показать описание

Lecture10.3 - Knowledge Graph Completion: TransE, TransR, DistMul, ComplEx
Jure Leskovec
Computer Science, PhD

Having talked about several knowledge graphs, we then introduce how to perform the knowledge graph completion task. In this lecture, we will talk about several models such as TransE, TransR, Dismult and ComplEx. For each of the model, we will discuss which type of relations it can model and relations it fails to model.

To follow along with the course schedule and syllabus, visit:
  1. Stanford Online
  2. Jure Leskovec
  3. Machine Learning
  4. Machine Learning with Graphs
  5. Graphs
  6. CS224W
Рекомендации по теме
Stanford CS224W: ML 0:10:31

Stanford CS224W: ML with Graphs | 2021 | Lecture 6.1 - Introduction to Graph Neural Networks

Stanford CS224W: ML 0:22:14

Stanford CS224W: ML with Graphs | 2021 | Lecture 13.1 - Community Detection in Networks

Stanford CS224W: Machine 0:20:27

Stanford CS224W: Machine Learning with Graphs | 2021 | Lecture 1.2 - Applications of Graph ML

Stanford CS224W: ML 0:20:10

Stanford CS224W: ML with Graphs | 2021 | Lecture 2.3 - Traditional Feature-based Methods: Graph

Stanford CS224W: ML 0:16:47

Stanford CS224W: ML with Graphs | 2021 | Lecture 2.2 - Traditional Feature-based Methods: Link

Stanford CS224W: ML 0:22:39

Stanford CS224W: ML with Graphs | 2021 | Lecture 16.4 - Robustness of Graph Neural Networks

Stanford CS224W: ML 0:34:31

Stanford CS224W: ML with Graphs | 2021 | Lecture 10.3 - Knowledge Graph Completion Algorithms

Stanford CS224W: ML 0:27:30

Stanford CS224W: ML with Graphs | 2021 | Lecture 2.1 - Traditional Feature-based Methods: Node

Stanford CS224W: ML 0:25:21

Stanford CS224W: ML with Graphs | 2021 | Lecture 9.1 - How Expressive are Graph Neural Networks

Stanford CS224W: Machine 0:11:55

Stanford CS224W: Machine Learning with Graphs | 2021 | Lecture 1.1 - Why Graphs

Stanford CS224W: ML 0:35:40

Stanford CS224W: ML with Graphs | 2021 | Lecture 12.1-Fast Neural Subgraph Matching & Counting

Stanford CS224W: ML 0:18:34

Stanford CS224W: ML with Graphs | 2021 | Lecture 5.1 - Message passing and Node Classification

Stanford CS224W: ML 0:20:07

Stanford CS224W: ML with Graphs | 2021 | Lecture 19.1 - Pre-Training Graph Neural Networks

Stanford CS224W: ML 0:24:52

Stanford CS224W: ML with Graphs | 2021 | Lecture 15.2 - Graph RNN: Generating Realistic Graphs

Stanford CS224W: ML 0:29:20

Stanford CS224W: ML with Graphs | 2021 | Lecture 5.2 - Relational and Iterative Classification

Stanford CS224W: ML 0:15:44

Stanford CS224W: ML with Graphs | 2021 | Lecture 15.1 - Deep Generative Models for Graphs

Stanford CS224W: ML 0:34:57

Stanford CS224W: ML with Graphs | 2021 | Lecture 10.1-Heterogeneous & Knowledge Graph Embedding

Stanford CS224W: Machine 0:16:50

Stanford CS224W: Machine Learning with Graphs | 2021 | Lecture 17.2 - GraphSAGE Neighbor Sampling

Stanford CS224W: ML 0:11:10

Stanford CS224W: ML with Graphs | 2021 | Lecture 16.1 - Limitations of Graph Neural Networks

Stanford CS224W: Machine 0:20:27

Stanford CS224W: Machine Learning with Graphs | 2021 | Lecture 1.3 - Choice of Graph Representation​...

Stanford CS224W: ML 0:12:41

Stanford CS224W: ML with Graphs | 2021 | Lecture 16.2 - Position-Aware Graph Neural Networks

Stanford CS224W: ML 0:31:52

Stanford CS224W: ML with Graphs | 2021 | Lecture 9.2 - Designing the Most Powerful GNNs

Stanford CS224W: ML 0:27:07

Stanford CS224W: ML with Graphs | 2021 | Lecture 3.2-Random Walk Approaches for Node Embeddings

Stanford CS224W: Machine 0:20:00

Stanford CS224W: Machine Learning with Graphs | 2021 | Lecture 14.2 - Erdos Renyi Random Graphs

Комментарии
Автор

Very cool concepts in this lecture. Thank you

איילתדמור
Автор

Thank you for lecture! I think the proof of that DistMul can't model inverse relations is not correct.

We have <h, r2, t> = <t, r1, h> for a specific h and t. Thus, the equation says that r1-r2 shoud belong to a specific hyperplane, however it doesn't mean r1 = r2. 26:51

BorisVasilevskiy
Автор

This lecture in particular was a great unified summary of methods! Thanks!

rherrmann
Автор

This is the greatest lecture of all time!

hohinng
Автор

i think i might be missing some points here but i'm still trying to understand how does textual/numerical information in KG is transformed before we use it with graph prediction model?

afrinaad
Автор

I don't understand the condition of complex conjugate of r_1 and r_2 at slide 65. Could you explain for me ? thanks prof

xuanloc
Автор

Can we use non-linear algorithms that are more expressive and more expansive that can capture all the relationships?

hanchisun