Retail
Machine & Deep Learning Optimization
SPAIN | 4 MONTHS
Key Technologies
ECOMMERCE
ANALYTICS
DEEP LEARNING
DATABRIKCS
AZURE
MACHINE & DEEP LEARNING OPTIMIZATION
01
THE CHALLENGE
Find a parallelization alternative for the training of neural models for massive data, seeking that the result is scalable and adaptive according to the amount and type of data that is processed.
myCloudDoor proposes the development of a training strategy that includes libraries such as Horovod and MirroredRunner, all this executed on a Databricks environment, for the creation of the model, Keras and Tensorflow libraries are used, which take advantage of the technology proposed by Azure.
02
THE SOLUTION
03
THE RESULT
A sample is created of how the different types of data should be approached, their transformations and strategies for processing and training in a distributed way so that the maximum benefit of Azure technology is obtained, thus achieving scalability and improvements in the times of data processing.
MACHINE & DEEP LEARNING OPTIMIZATION
01
THE CHALLENGE
Find a parallelization alternative for the training of neural models for massive data, seeking that the result is scalable and adaptive according to the amount and type of data that is processed.
02
THE SOLUTION
myCloudDoor proposes the development of a training strategy that includes libraries such as Horovod and MirroredRunner, all this executed on a Databricks environment, for the creation of the model, Keras and Tensorflow libraries are used, which take advantage of the technology proposed by Azure.
03
THE RESULT
A sample is created of how the different types of data should be approached, their transformations and strategies for processing and training in a distributed way so that the maximum benefit of Azure technology is obtained, thus achieving scalability and improvements in the times of data processing.