Session starts - 14:00
Implementing Streaming Decision Tree Using Approximative Algorithms in Flink
Much data, including the ones used for learning phase in a machine learning system, come in a set of streams. The use of streaming machine learning algorithms then offers several advantages over the batch ones. The talk explains an attempt made in Amadeus Travel Intelligence to use Flink to implement a streaming decision trees algorithm.
About the speaker
Anwar Rizal is a research engineer in Amadeus. He works in Travel Intelligence business unit, a dedicated business unit in Amadeus to handle the travel data. Streaming machine learning and understanding travellers are two of his favorite subjects.