Data Reply IT
I am a data engineer and technical head of data governance department at Data Reply IT. My focus are realtime data solutions, distributed computing and GDPR compliance solutions. My experience ranges from realtime data ingestion and analysis solution for big data architectures to neural networks training optimization on distributed infrastructures. I'm working on GDPR compliance designs and solutions too.
Speeches di Andrea D'Amelio
Distributed training of the Inception deep learning architecture through Spark and Deeplearning4j
Deep architectures enable us to reach unbelievable accuracy values, but what's the cost in terms of computational effort and time? Find out the structure of the Inception architecture, feeding the state of the art of the best algorithms of image processing, and how Spark and DL4J can efficiently train complex neural nets by distributing operations over multiple GPUs. Convoluted models requires massive data processing, so speed really matters. Well show how faster hardware, optimized code, and data parallelism empower developers to tackle the problem.Lingua speech: Italian
Sensors and Data Acquisition, Machine Learning and Artificial Intelligence, Algorithms
Real Time Big Data Processing Solution in a Leader Insurance Company
Collecting and analyzing data in the shortest time lapse possible is the main challenge when you're dealing with big data, particularly when you have the responsibility to help people.
The speech aim is to show an end-to-end solution based on Kafka and Spark Streaming that allows the real-time ingestion and processing of data coming from telematics devices installed on our cars (but also bikes and pets), in order to give assistance to the customers.
The solution enabes the client to make more fast and precise decisions, thanks to a fault tolerant architecture based on Hadoop technologies.