NOKIA Research Center, Eurolaboratory, Cambridge (UK)
Jani Kivioja is a Research Leader at Nokia Research Centre Cambridge. The focus of research in the Cambridge unit is around nanotechnology. Kivioja’s team has been actively working on nanowire sensing in order to develop an artificial nose capable of sensing different substances. You can learn a more about nanowire sensing by reading this Cambridge news article or by watching this nano minute video.
Listen to the full interview:
Summary of the interview:
Could you tell me about your area of research?
My research is related to nano-sensing architectures.
We are trying to find new ways of integrating sensors, gathering the data and building totally new kinds of nanotechnology enabled sensors. All these sensors are somehow related to sensing environmental parameters, which can be for instance gases or chemicals.
What is the talk topic “Nanonose: Machine learning with nanowire sensors” about?
In this researcg we used semi-conducting nanowires to measure different gases or solvents.
We are utilizing a large variety of different parameters. The response in different sensors is always somewhat different, although you are using nanostructures for sensing purposes. The idea is that we use machine learning algorithms to take advantage of the variety of parameters coming from different sensors. By telling computers what material is being sensed, the next time it can recognize it, even without us having to know the physical phenomena behind the materials. The phenomenon just has to be reproducible. The idea is that the system is scalable in terms of number of sensors.
The overall aim is to help mobile device users in their everyday life, so that they can use their device to detect, for instance, whether food is fresh or bad.
How your research can be applied in industry?
Besides mobile devices, one obvious application is quality control in the food industry. Of course, the same approach for handling the data could be applied to any other network of sensors.
Instead of trying to understand why something is responding in certain way, we can just teach a system with a large amount of sensors that a particular response means something.
