Solution Idea#
Forecast
According to Notes from the AI frontier, McKinsey&Company, time-series data has the second highest number of business use cases. And time-series forecasting is a common statistical task in business, where it helps to inform decisions about the scheduling of production, transportation and personnel, and provides a guide to long-term strategic planning.
Explainability and Simulation
Support ante-hoc explainability, such as cause finding, key driver identification. Once we know the physical meaning of each variable, we are able to test numerous strategy simulations overtime before implementation. This analysis gradually progresses from understanding “what” to uncovering “why” and finally determining “how”.
Decision making
Sequential decision-making (aka. planning) is the problem of computing action strategies for a robot to achieve good long-term returns when actions may have long-term consequences. Many robotics problems fit the above planning in non-deterministic and partially observable scenarios, such as human-robot collaboration, homecare, and underwater navigation. The ability to compute reliable and robust decisions in the presence of uncertainty is essential in modern autonomous system.