Zhen Li is an AI Scientist with a strong background in time series analysis and optimization techniques. After crossing the ocean from the US to join Silo.AI in Finland, he has been applying machine learning and putting his Python skills to use at deploying AI solutions for clients in process industry, manufacturing and finance, to mention a few. One of his industrial projects include a water quality prediction model for Ramboll.
Time series today, reinforcement learning tomorrow
Originally from China, Zhen started his academic career studying Marine Science at Xiamen University. After that he went on to the US to pursue a PhD in Computational Chemistry. His background gives him a good position to work on machine learning related to chemistry, pharmaceutics and mathematics, but also to jump into process and optimization driven tasks.
At Silo.AI, Zhen has been involved in many projects with time series data. Zhen especially finds time series projects interesting as they present a difficult challenge of predicting the future:
“It’s exciting to look at the historical curves, as history is surprisingly repetitive. Also, we need to be aware that prediction is very difficult, especially when it is about the future. It’s a great moment when you see good results with time series data in our client projects”, Zhen says.
Reinforcement learning (RL) is another area that excites Zhen in his field. Reinforcement learning is a type of machine learning where the model is not programmed to do something, nor is it learning from labels. Instead, it learns in a similar way people learn: by getting a reward or a punishment.
Zhen sees reinforcement learning as an exciting technology for the future, that has already shown great progress in the gaming world. As reinforcement learning requires a vast amount of “learning times” (the basic RL agents’ way of learning is trial and error) it has the most potential of being used in simulated environments, such as in digital twins.
“Reinforcement learning is the closest method to artificial intelligence, because of similar learning mechanism as we humans have. Other methods of machine learning can be thought of as an extension to statistics”, Zhen explains.
It’s the outcome that matters
Zhen enjoys creating useful solutions to the client, where he can see clear value and how that improves the clients’ business. He likes to put his interdisciplinary background into use at clients from different industries and describes his thinking as innovative. Zhen is at ease in teams with people from different backgrounds and varying levels of technical or mathematical skills.
Sports and travelling
Zhen hopes that during his free time he can be found at the gym. He likes to organize and spend time on road trips and enjoys an occasional console game.
Favorite Silo.AI value?
“My favorite Silo.AI value is Build Bonds. I like my colleagues here and believe that power of team work is more than the sum of individuals. I like the communication and inspiration from my colleagues, and I enjoy working closely with peers on projects to tackle Mission Impossible.”