Corpus ID: 49310354

Canonical Tensor Decomposition for Knowledge Base Completion

@inproceedings{Lacroix2018CanonicalTD,
  title={Canonical Tensor Decomposition for Knowledge Base Completion},
  author={Timoth{\'e}e Lacroix and Nicolas Usunier and G. Obozinski},
  booktitle={ICML},
  year={2018}
}
The problem of Knowledge Base Completion can be framed as a 3rd-order binary tensor completion problem. In this light, the Canonical Tensor Decomposition (CP) (Hitchcock, 1927) seems like a natural solution; however, current implementations of CP on standard Knowledge Base Completion benchmarks are lagging behind their competitors. In this work, we attempt to understand the limits of CP for knowledge base completion. First, we motivate and test a novel regularizer, based on tensor nuclear $p… Expand

Figures, Tables, and Topics from this paper

Binarized Knowledge Graph Embeddings
TuckER: Tensor Factorization for Knowledge Graph Completion
Tensor Decompositions for temporal knowledge base completion
Tucker decomposition-based Temporal Knowledge Graph Completion
Generalizing Tensor Decomposition for N-ary Relational Knowledge Bases
TEMPORAL KNOWLEDGE BASE COMPLETION
Convolutional Complex Knowledge Graph Embeddings
...
1
2
3
4
5
...

References

SHOWING 1-10 OF 40 REFERENCES
On Tensor Completion via Nuclear Norm Minimization
Complex Embeddings for Simple Link Prediction
Scalable and Sound Low-Rank Tensor Learning
A Three-Way Model for Collective Learning on Multi-Relational Data
Multitask learning meets tensor factorization: task imputation via convex optimization
Combining Two And Three-Way Embeddings Models for Link Prediction in Knowledge Bases
Learning Structured Embeddings of Knowledge Bases
Convolutional 2D Knowledge Graph Embeddings
...
1
2
3
4
...