Pushing the technological frontiers in the battle against food crime, this project aims to implement an accurate yet efficient, cost-effective and easily deployable quantitative solution for food fraud detection and the screening of agri-food samples which combines the use of portable sensors and machine learning. The adulteration of food is one of the fastest growing economic crimes; the deliciously deceptive foods we eat may be packed with things that are not supposed to be there—a practice known as “food fraud”. Spectroscopy is a very suitable means for food quality control as this technology can identify the unique “fingerprint” in agri-food products. It offers non-destructive ways to counter fraud in industry, in the food supply chain, and to ensure public safety. Field deployment of high resolution spectrometers and remote monitoring is a major priority yet remains elusive due to the very high cost of spectroscopy devices which are also relatively large and cumbersome.
On the other hand, miniaturised spectrometers can solve the issue of portability but they can only attain a limited performance and operate on a limited wavelength range . Machine learning is finding increasingly broad applications in food sciences. This most often involves building a model relationship between a dependent, measurable output and an associated set of controllable, though complicated, independent inputs.
Techniques used address reducing the dimensionality of input data and use both unsupervised & supervised learning approaches. They can include the common holistic regression-based partial least squares model which is a standard tool in chemometrics, principal component analysis, local features based solutions and deep learning. Augmenting the performance of portable spectrometers can be well justified and argued. It can extend the limited range of individual sensors to cover near infra-red, visible and ultra violet wavelength range through fusion. It is otherwise very useful in addressing the intrinsic spectral data curse of dimensionality, non-linearity and collinearity where a single peak at a given wavelength may bleed into the measurement of multiple neighbouring data points. Further, previous work has shown that spectral data can be characterized by relatively low signal-to-noise ratio, it is of low resolution, and may be affected by background and ambient noise including EM events (such as switching lights on and off).
The misalignment of spectral data is yet another common distortion. All these challenges have to be addressed and embedded in any machine learning approach so that meaningful information is not discarded due to the low resolution of portable spectrometers . It seems that putting back ordinary citizens and users of the portable spectrometers technology to the first line of defence against food fraud rather than being the victims depends largely on addressing the points above. Data acquired using portable sensors will be processed and analysed using machine learning algorithms for the purpose of quality control.
Hence, this project will have the following objectives:
1)To build statistical models which tackle multivariate spectral data sets.
2)To acquire and adequately use spectral data for deep neural network training.
3)To identify and extract local features which are distortions invariants and suitable for matching spectral data.
Vice Chancellors Research Scholarships (VCRS)
The scholarships will cover tuition fees and a maintenance award of £14,777 per annum for three years (subject to satisfactory academic performance). Applications are invited from UK, European Union and overseas students.
The scholarship will cover tuition fees at the Home rate and a maintenance allowance of £ 14,777 per annum for three years. EU applicants will only be eligible for the fees component of the studentship (no maintenance award is provided). For Non EU nationals the candidate must be "settled" in the UK.
As Senior Engineering Manager of Analytics at Seagate Technology I utilise the learning from my PhD ever day
Adrian Johnston - PhD in InformaticsWatch Video
Monday 18 February 2019
mid March 2019
The largest of Ulster's campuses
When applying for this PhD opportunity please quote reference number: