Machine learning models must keep up with the changing world

Life Science Informatics Professor Indrė Žliobaitė develops computational methods that are explainable, trustworthy, and robust to change. With these methods, it is possible to analyse and understand the changing world on a large scale.

What are your research topics?

My research has two main directions: developing computational methods that are explainable, trustworthy, and robust to change, as well as using those methods to analyse and understand the changing world on a large scale.

My background is in adaptive machine learning. As the world changes, sooner or later earlier models go out of tune. Adaptive machine learning is about keeping models robust, accurate and up to date. The analytical side of my research involves analysing large-scale patterns of change in nature and society over time, using tailored computational methods that we develop.

Where and how does your research have an impact?

The main impact of my research is in knowledge discovery that helps to better understand ongoing change processes in nature and society. 

Practical uses of my research are in strategic planning, particularly in better understanding the structure of changes, and how fast and how often things change.

My research also helps to better understand what is possible to predict and what is not. We work with scientists, municipal and governmental agencies and consultants.

The main application areas of my research include:

  1. Understanding biospheric and social change processes over short time scales, such as the effects of the recent pandemic, and over long time scales, such as the structure of biospheric change over hundreds, thousands, and millions of years, as well as ecological baselines for the future.
     
  2. Understanding longevity of entities, including longevity and ageing of biological species, industries and cultural phenomena, such as music genres or fashion trends.
     
  3. Helping scientists with scientific reasoning that includes the use of modern machine learning methods and AI in general. Characterizing trustworthiness, transparency and resilience of AI for scientific reasoning.

What is particularly inspiring in your field right now?

The rapid and broad adoption of AI in the society is inspiring. In the last decade AI expanded from a specialist activity to broad use in many circumstances of our everyday lives. Ethics, philosophy and safety of AI are particularly inspiring topics for the years to come.