원문정보
초록
영어
This paper discusses the motivations and principles of Deep Boltzmann Machines regarding learning algorithms for deep architectures. Policy-based systems management (PBM) and Deep Boltzmann Machines (DBM) are two of the many techniques available for artificial intelligence (AI), each having specific benefits and limitations, and thus different applicability; choosing the most appropriate technique is the first of many challenges faced by the developer. The discussion forms a backdrop for a detailed evaluation of the two techniques, in which the concepts underpinning each of PBM and EBM are reviewed and placed into context with each other as well as with the other popular techniques for AI. After considering the operation and suitability of the techniques in isolation, the focus shifts to look at how PBM and DBM could be combined in complementary ways to achieve more sophisticated and versatile AI systems for E-commerce..
목차
1. Introduction
2. BackgrounD: RBMS and their Generalizations
3. Multimodal Deep Boltzmann Machine
3.1. Salient Features
3.2. Modeling Tasks
4. In Search of Synergies for E-Commerce
4.1. Dataset and Feature Extraction
4.2. Model
4.3. Classification
4.4. Quantification
5. Conclusion
Acknowledgment
References