Distributed Architectures for Machine Learning
Event details
| Date | 01.07.2016 |
| Hour | 12:00 › 14:00 |
| Speaker | Georgios Damaskinos |
| Location | |
| Category | Conferences - Seminars |
EDIC Candidacy Exam
Exam President: Prof. Karl Aberer
Thesis Director: Prof. Rachid Guerraoui
Co-examiner: Prof. Boi Faltings
Background papers:
Tencent Rec: Real-time Stream Recommendation in
Practice
HyRec: Leveraging Browsers for Scalable
Recommenders
Distributed Asynchronous online learning for natural language processing
Abstract
The big data era highlights a major scalability challenge for online machine learning. Addressing this challenge requires a holistic approach that involves distributed solutions both from the algorithmic and from the system perspective. In this proposal, we first study two scalable recommender systems with different architectures along with multiple machine learning algorithms. TENCENTREC is a real-time scalable recommender with a centralized framework. HYREC is an online recommender that employs a hybrid architecture. We then study an asynchronous algorithm with promising characteristics towards designing scalable online machine learning algorithms. In this proposal, we argue that a democratized machine learning system must enable all service providers to deploy their service and all clients to use it regardless of their resources. Finally, we examine recent work concerning general purpose large-scale machine learning frameworks and propose future work towards democratizing machine learning.
Exam President: Prof. Karl Aberer
Thesis Director: Prof. Rachid Guerraoui
Co-examiner: Prof. Boi Faltings
Background papers:
Tencent Rec: Real-time Stream Recommendation in
Practice
HyRec: Leveraging Browsers for Scalable
Recommenders
Distributed Asynchronous online learning for natural language processing
Abstract
The big data era highlights a major scalability challenge for online machine learning. Addressing this challenge requires a holistic approach that involves distributed solutions both from the algorithmic and from the system perspective. In this proposal, we first study two scalable recommender systems with different architectures along with multiple machine learning algorithms. TENCENTREC is a real-time scalable recommender with a centralized framework. HYREC is an online recommender that employs a hybrid architecture. We then study an asynchronous algorithm with promising characteristics towards designing scalable online machine learning algorithms. In this proposal, we argue that a democratized machine learning system must enable all service providers to deploy their service and all clients to use it regardless of their resources. Finally, we examine recent work concerning general purpose large-scale machine learning frameworks and propose future work towards democratizing machine learning.
Practical information
- General public
- Free
Contact
- Cecilia Chapuis EDIC