Abstract | ||
---|---|---|
In this paper we aim at facilitating generalization for deep networks while supporting interpretability of the learned representations. Towards this goal, we propose a clustering based regularization that encourages parsimonious representations. Our k-means style objective is easy to optimize and flexible, supporting various forms of clustering, such as sample clustering, spatial clustering, as well as co-clustering. We demonstrate the effectiveness of our approach on the tasks of unsupervised learning, classification, fine grained categorization, and zero-shot learning. |
Year | Venue | Field |
---|---|---|
2016 | ADVANCES IN NEURAL INFORMATION PROCESSING SYSTEMS 29 (NIPS 2016) | Categorization,Interpretability,Computer science,Unsupervised learning,Regularization (mathematics),Artificial intelligence,Conceptual clustering,Cluster analysis,Machine learning |
DocType | Volume | ISSN |
Conference | 29 | 1049-5258 |
Citations | PageRank | References |
4 | 0.39 | 0 |
Authors | ||
4 |
Name | Order | Citations | PageRank |
---|---|---|---|
Renjie Liao | 1 | 204 | 13.34 |
Alexander G. Schwing | 2 | 696 | 51.78 |
Richard S. Zemel | 3 | 4958 | 425.68 |
Raquel Urtasun | 4 | 6810 | 304.97 |