Faglige nøgleord: Nanotechnology, machine learning
Oplæg tilgængeligt på: Engelsk
About me:
I am a curious and a bit goofy PhD student from China, and have been studying in Denmark since high school. I am a firm believer in that complicated science can be and should be communicated in an easily-accessible way with the mass audience, and I would love to show the possibilities to the young people nowadays!
About the research:
Under the broad field of nanotechnology, we study and characterize nano-scale materials using electron microscope, that has resolution down to the atomic scale (you can see individual atom!). For my research, I program and automate process to collect and analyze data, and use machine learning to infer their possible structures.
About the presentation:
As the title suggests, we will play as the "detective" to "catch" the "nano-killers" in our daily life. As industry advances, more and more harmful and toxic nanoparticles are being released into the air without even being detected or known. Luckily, some of these particles have distinct crystal structure, which produces distinct patterns under the electron exposure. The presentation will draw analogy between:
-
Finding the criminal using fingerprints.
-
Finding the particle structure using electron diffraction pattern.
We will start with handing out some printouts of fingerprints, and ask them to match against a dozen candidates. Then we will hand out some printouts of electron diffraction patterns and ask them to match as well. This will get progressively harder as we start to cut out parts of the fingerprints, smear/rotate them, as well as with added noise on the images.
This is to illustrate the difficulty and the need for proper image treatment. We will discuss a few algorithms in its basic ideas and their applications. Then we will talk about machine learning and why they are particularly useful in the task.
About the sustainability:
The project aims at having good impact on SDG 3, 4 and 9, by bypassing the lack in accuracy using cheaper result tools with the aid of machine learning, and hence encouraging more research on airborne nanoparticles and other nanomaterials that influence our health.