
Case History
Bank

REINFORCEMENT LEARNING FOR FINANCE
Development of Advanced Analytics applications
Requirements
Use GPUs for neural network training techniques
Solution
Development of a flexible environment to support the transition of the project to an AI environment
APPLICATION
Machine Learning and Deep Learning
KNOW-HOW
Technical skills, Training, POC
ADVANTAGE
Process acceleration, multiuser environment, user friendly interface
Our customer is a leading bank in Italy and is experimenting with reinforcement learning techniques with artificial intelligence in the area of trading. In a first phase, it has tried to train a neural network to be able to make correct investment selections in the market with the aim of providing support to the physical trader, especially useful in moments of panic when a human trader could make wrong decisions. The customer requested the use of GPUs for the development of the project.
E4 provided operational and strategic support for the implementation of a system able to best use the potential of GAIA. For this, it proposed an approach based on servers with two GPUs and several days of training. It then implemented a multiuser environment to be used by the internal development team so that each developer would have a Jupyter Notebook interface to work with. The development environment thus becomes a container system able to ensure the availability of a work environment over time to be able to reproduce processes and results already obtained in the past.
The customer appreciated our know-how in the world of AI and our advanced Python development techniques. To this, we added extreme flexibility in the planning and development of technical solutions.