We have just said goodbye to the wonderful year 2017. Wish you all had a fulfilled Christmas holiday! Opportunities and discoveries are waiting for us in the year 2018.
After coming back to academia from industry for around 4 months already, I have noticed some differences between being a software engineer in a company and a researcher pursuing a PhD degree in an university. The most significant difference is that I could just implement the novel idea and write a documentation to explain about the innovation that I found in a company, while it is necessary to design scientific methods to best prove the idea is not wrong from all different angles in the academia. Another big difference is that time management is more flexible in academia, which on one hand is a great advantage if the time can be managed well and a challenge on the other hand. Therefore, learning and training on scientific thinking and time management will be focused in the upcoming months.
Since my last blog, my work during this period of time has mainly focused on carrying out literature review, proposing general research goals for the upcoming three years, scheduling more detailed plans for the first year research, and testing out the state-of-the-art image analysis techniques on the smartphone-and-cloud-services-based test strip reading system that we have developed. The difficult part during the research proposal was that apart from maximising potential collaboration with the other ESRs in the FoodSmartphone project, the proposed research objectives should also be set within the scope of image analysis and satisfy the requirements of a PhD project at the same time. Thankfully, this difficulty has been solved. In the near future, I will explore more on the fields of optics, biochemical assays, food analysis, and my background fields including computer science and network engineering. I believe the imaging systems that we are about to develop in the upcoming three years will make a difference for on-site food analysis.
In order to make this blog more interesting to read, I would like to introduce a little about image segmentation, which is one of the essential sub-fields in image analysis. Image segmentation techniques based on machine learning can be broadly divided into three categories: unsupervised segmentation techniques (e.g. K-means clustering) require prior knowledge about the dataset. An advantage of using unsupervised segmentation methods is that they do not require labelled images for training. In contrast, supervised methods (e.g. Artificial Neural Networks) require a lot of labelled images that are costly to prepare. However, current neural network related techniques outperform the others in terms of segmentation accuracy, and some of them even have surpassed the segmentation accuracy achieved by human in some scenarios. Therefore, supervised techniques have gained tremendous research attentions during the last few years. Another machine learning branch other than supervised and unsupervised learning is the reinforcement learning. Reinforcement learning techniques require very few labelled data which is called “reward”, and they will explore the rest of unlabelled data by themselves. Unsupervised and reinforcement learning are the two primary mechanisms working within our human visual system, and the state-of-the-art image segmentation techniques using these two types of learning are still away from the accuracy achieved by human. Therefore, Unsupervised and reinforcement learning based image segmentation techniques are promising for us to explore.
One of the main subjects of my daily life being a researcher at Queen’s is to feed myself. Exploring tasty local food in restaurants is always joyful. Some of these local food are innovative. For instance, the first photo below shows the BAOs which I call it “Asian hamburgers”. They are made by steamed bun covering meat inside with tops above, and they are delicious! One of the main focuses during eating these food is how not to mass up my face or the table, and I think this is going to be one of the challenges to be solved during my PhD. Other than taking food exploitation around Queen’s University Belfast, I have also started cooking by myself at home. At the beginning, my cooking methods are limited in boiling, steaming, and frying. I could make a couple of dishes using common food materials that can be found in the supermarket such as cod, salmon, broccoli, and carrot. By the time goes, my cooking methods got extended to baking as well. And baked beef was added into my menu recently. Some of my masterpieces are shown in the photos below 😛
I spent a decent Christmas holiday at home with my families. We enjoyed some traditional Chinese food, for instance, the food shown in the second photo below at the top right is called “Zongzi” or rice-pudding with lotus leaf covering the outside. We also revisited some of the famous sites in the city Hangzhou.
The year 2018 is here, and the year of dog is coming. In the end of this blog, I hope you all are doing well, and wish you a fulfilling year 2018 and to be a lucky dog in the upcoming year of dog!