GraphLab brings together the leading experts in machine learning, graph analytics, and scalable system design. We are located in Seattle and funded by a world-class investors. Most importantly, we are a team that is committed to helping people realize the full potential of their data.
Grokking Graphs and Machine Learning
The GraphLab platform combines a high-level graph-parallel API with sophisticated graph computation engines. By abstracting away the challenges of scalable multicore and distributed systems, GraphLab simplifies the design and implementation of large-scale machine learning algorithms.
GraphLab enables state-of-the-art scalable graph computation by using:
- asynchronous low-latency communication
- intelligent graph partitioning
- sparse representations
- dynamic scheduling
- fast user-mode threading
- scalable concurrent data-structures
What’s available in GraphLab today?
The toolkits are a collection of console tools that easily scale from personal workstations to cloud instances:
- ALS and SGD for recommender systems
- k-means++ for clustering data
- LDA for modeling collections of topics
- MRF and CRF inference algorithms for structured prediction tasks
- Label propagation for fraud detection
- PageRank and triangle counting for graph analytics
- Community detection for social network analysis.
Many users also develop their own algorithms by using the GraphLab C++ API directly, while Python and Java APIs are on their way.
Get started with GraphLab!
On graphlab.org you can find out how to quickly install GraphLab, whether it’s your own laptop, a Hadoop cluster, or just some cloud instances. We also have tutorials for applying the above algorithms to your own data.
Join the community!
Sharing innovation is a core part of our culture and GraphLab will remain an open-source project with code available on Github. As a company we will support our customers by developing new technology and toolkits on top of the GraphLab platform. We will continue to actively reach out to our community to guide GraphLab’s future, both promoting and incorporating tools that the community builds.