Svitlana Galeshchuk - Artificial intelligence for Prediction in Financial Market

Organized by: 

Dr. Svitlana Galeshchuk has earned her Master degree in Management of European Affairs from Lille 1 University - Science and Technology in France, and her doctorate in Economics from Ternopil National Economic University, Ukraine.
She has served a Fulbright Scholar visiting the College of Engineering and Computing at Nova Southeastern University, the USA. She has a faculty position at Ternopil National Economic University.
Dr. Galeshchuk is currently acting as a Visiting Professor at the Laboratory of Informatics (LIG, MAGMA team), Université Grenoble Alpes. Her research interests are in the areas of market inefficiencies and economic forecasting. She is currently investigating the relative merits of econometric and artificial intelligence for technical and fundamental analysis in foreign exchange markets. Her focus is in the area of neural computing.
Current market conditions are in the process of modification. It is caused by technological innovations, modern approaches in management and finance, new financial instruments, transformation of the role of central banks and government. We believe, apart from the standard set of econometric technics which are losing their effectiveness now, the new mainstream in the analysis of economic systems and forecasting of financial markets’ dynamics is emerging. This mainstream is based on the application of artificial intelligence technics to predict economic indices. Thus, Dr. Galeshchuk will focus on the development and successful application of these innovative scientific approaches to nonlinear analysis and prediction of financial markets with cutting-edge deep learning methods and agent-based systems.
Over the last decade brain inspired deep learning technologies has been proven to be a very robust and effective prediction method in a variety of application domains. Deep networks have had success in time series forecasting applications and have also been used for financial predictions. The recent success of deep networks is partially attributable to their ability to learn abstract features from raw data. Dr. Galeshchuk will target empirical studies that prove deep neural networks are significantly better at financial prediction than existing time series models. In her presentation Dr. Galeshchuk will identify questions raised by recent her work on foreign exchange prediction that appears in peer-reviewed journals. She will also provide brief overview of deep learning architecture types and relevant programming environment (i.e., Tensorflow, Caffe, Matlab).
Agent-based modelling is a powerful technology to address high complexity problems in a decentralized way. It has led to the new branch of Computational Economics. Dr. Svitlana Galeshchuk will introduce the agent-based modelling methodology and emphasize its application to simulate and predict currency markets. Design problems occurring when creating an agent-based financial market will be discussed. Existing agent-based frameworks for market simulation will be presented.