Lynx will speak about its pioneering approach to utilize communicating neural networks to predict missing attributes of graphs. In plain business language, data on individuals and their relationships (e.g. the subscribers of a telecom/banking service and the links between them) can be represented as a graph. In such graphs, individuals are represented as nodes, which can have attributes such as age, gender, hobbies, location, service usage behavior.
This type of information is usually far from complete, therefore, complicating tasks that require the segmentation and targeting of individuals. Lynx came up with a groundbreaking approach, applicable to real use cases, to overcome this problem which will be shared during this talk.
For more information, go to: Spark Summit East 2017