Memory-Distributed Singular value decomposition.
- Mentors
- Daniel Coquelin
- Organization
- Forschungszentrum Jülich
- Technologies
- numpy, pytorch, Intermediate/advanced programming, parallel/distributed computing
- Topics
- machine learning, data science, linear algebra, To fill the gap between Data analytics and Machine learning libraries
The major goal of the project is to develop a distributed SVD algorithm that is both efficient and numerically stable in Heat. This will be a major boost as the number of applications of the SVD algorithm is high, In most of the applications basic principle of Dimensionality Reduction is used.
Applications of SVD algorithm are: Image Compression, For recognition of faces, Removing Background from Videos, and Finally, the SVD algorithm is also the backbone of recommender systems such as Amazon, YouTube, Netflix, and many others.