Home » Gadgets » About Michelangelo: The Machine Learning in the Background

About Michelangelo: The Machine Learning in the Background

Thursday, September 7th, 2017 | Gadgets

The system behind the service: the American company Uber has presented its own machine-learning platform, Michelangelo, which since its integration in 2015 takes over many functions. For example, the system is used in the Uberats app to calculate delivery times for food.

For this purpose values ​​such as the scope of the order, the utilization of the restaurant and the general reliability of the service provider are compared with customer evaluations. Already this process of food delivery involves several factors. Therefore, the calculation of the estimated delivery time is not a simple matter.
Michelangelo was developed from a mixture of open-source systems and specially created algorithms. The source components include HDFS, Spark, Samza, Cassandra and Mllib. The software has been specially tailored to the Uber infrastructure. It has access to all stored data of the company and its customers.
The program is designed to cope with six work steps. In a fixed sequence, managing data, training ML models, assessing these models, setting up and executing and recording estimates.
From a decentralized model to enterprise-wide software
Until the start of Michelangelo, according to Uber many data analyzes had been implemented decentrally in the individual company departments. Partially, employees developed their own search algorithms on the local desktop computer. Accordingly, the computational resources and performance were limited. Furthermore, results could not be stored centrally and company-wide.

        
                
            
    Michelangelo works in several steps. (Picture: Uber)

The rapid growth of overstrained a better solution that resulted in Michelangelo. The software standardizes workflows and tools for all employees of the company. Data is stored centrally and is accessible from anywhere. This should also speed up new projects.
Services that offer surplus services are not available in Germany. The European Court of Justice (ECJ) has classified the company as a tax service, subject to many conditions. The business model would not be profitable with these restrictions.

Related

Mambo FPV: Parrot announces FPV drone fo

The French manufacturer Parrot has presented an FPV version of its Mambo drone.

Bachelorette Finale Is When

Bachelorette Finale Is When Bachelorette Finale Is When Vlogs 'The Bachelorette: Men Tell

Pala Investments - Fund from Zug relies

The fund wants to buy metals used in electric cars: Pala New Energy

Chris Cornell Death

Chris Cornell Death Chris Cornell Death Vlogs Pearl Jam – Breath Chris Cornell,