Simulation models to guide AI learning - overview
Abstract:
AI has been hugely successful across many industries by exploiting the huge volumes of data being generated today. Some domains, however, do not have sufficient data volumes to fully exploit the capabilities of AI. A common example is forecasting physical systems (e.g. weather, ocean circulation), where sensor data are often sparse but the system dynamics is well understood via physical rules generally encoded via a set of partial differential equations. This work explores the viability of using simulation models to guide the training of AI systems
Required Skills:
- Working knowledge in deep neural networks and applications such as regression or classification
- Strong Python programming skills
- Knowledge of Generative Adversarial Networks preferable