DataTrading programmers are mentoring at Microsoft Conference
Last Thursday our Head Data Scientist – Bohdan was a speaker at the Microsoft conference in Unit city.
The presentation consisted of 2 parts. The first was devoted to the architectual peculiarities of the networks. Bohdan demonstrated how to create basic networks, told about the RNN, LSTM networks’ architectures and reviewed their components in more details. He also compared the FFN, RNN, LSTM in action. He taught his audience how to teach the networks to predict the amount of the solar energy (approximately the same models are used while trading forecasting).
Furthermore, Bohdan demonstrated the importance of understanding the gradient descent methods and the training process. Two methods that don’t have much differences, but produce such opposite results. Later on, he gave some pieces of advice on training neural networks and hardware. Of course, there’re plenty questions from the audience and a brief discussion.
We are proud that you are a part of our dream team!
Check out Bohdan’s interview!