Research areas

While there has been much progress in machine learning, there are also challenges:

  • the computational complexity of machine learning algorithms is in most real-life applications very high. To tackle this challenge, we work towards designing lightweight algorithms or implementations.
  • the mainstream machine learning technologies are black-box approaches, making us concerned about their potential risks. To tackle this challenge, we work towards making machine learning more explainable and controllable.

In many domains such as agronomy, (agro)chemistry, (plant/animal)biology, and hydrometeorology, people usually seek elegantly simple equations to uncover the underlying laws behind various phenomena. The question in the field of machine learning is: Can we reveal simple laws instead of designing more complex models for data fitting?

Although there are many challenges, we are still very optimistic about the future of machine learning, especially targeting problems in the agri-food value chain. As we look forward to the future, here are the research hotspots we are focusing on.

Scalable Machine Learning & Edge Computing

With the rise of the Internet of Things (IoT) and the widespread use of AI, the combination of machine learning and edge computing to analyze and process data near the data generation source, has become particularly important.

Quantum Machine Learning

Quantum machine learning is an emerging interdisciplinary research area at the intersection of quantum computing and machine learning. Quantum algorithms have surpassed the best classical algorithms in several problems.

Explainable Machine Learning

The ability gap between machine and human on complex cognitive tasks becomes narrower and narrower. However, we are still in an early stage in terms of explaining why those effective models work and how they work.

Improvisational Learning

Improvisational learning, assumes that the world is full of exceptions. Intuitively, the system acquires knowledge and problem-solving abilities via proactive observations, instead of being optimized towards a preset goal.