6 September 2022,

by Katarina Alexander, Edinburgh Napier University

This year’s IRDTA Deep Learn Summer School took place in Gran Canaria from July 25^th – 29^th. The summer school was both in-person and online, with the majority of the nearly 500 participants coming to the island. The three keynote speakers – Wahid Bhimji, Joachim M. Buhmann and Kate Saenko, were brilliant and really brought to light their areas of expertise. Subjects covered by the summer school included Explainable AI, Natural Language Processing, Computer Vision, AI in the Natural Sciences, Deep Learning Design and more.

Speakers gave short four-and-a-half-hour-long course on their subject split into three lectures across two days, with three speakers in parallel so we could attend courses that were connected to our own research. The lectures were also recorded, in case of overlap and lecture materials available to download, something I have taken advantage of. For many of the participants I met this was the first conference we were able to attend in person, and it was great to meet with researchers from across the globe. The speakers were excellent and managed to fit a lot of information into such a relatively short time. One of the benefits of the hybrid delivery as a person with a disability was being able to attend from my hotel room on a day I felt unable to attend in person, I think this approach works well from an accessibility stand-point.

Overall, the IRDTA schools are fantastic, and I would definitely encourage anyone to attend if you have an interest in any form of deep learning! Details of their Autumn 2022 and Winter 2023 schools are available at https://irdta.eu/#active-events.

 

5 September 2022,

by Dr Juraj Sikra, University of Strathclyde

“The limits of my language mean the limits of my world.” Wittgenstein

On my way to Amsterdam, I was thinking what the research into improving cybercrime reporting really needs. I thought that to do research that convinces the government to take cybercrime more seriously, I must be able to tell the story of cybercrime victims and those who help them in the simplest format whilst preserving the drama that takes place in people’s lives.

This was confirmed to me during the informal discussions with Dr Rutger Leukfeldt when we debated how cybercrime research is evolving in the Netherlands. He spoke about the tensions of doing government research in this area, which favours funding projects in traditional crime (i.e., robberies vs. online scams). I believe that this is, in part, because we’re still only learning to tell the story of what happens in cybercrime. As Wittgenstein theorised, language defines the boundaries of reality. Unless we learn to speak about something effectively, it is as if it didn’t exist. Dr Leukfeldt observed that research into SMEs (Small-to-medium sized enterprises) funded by government proxies is flourishing. I wondered whether the Dutch government fails to see the tax-paying cybercrime victim as its primary stakeholder but takes the reverse approach to the Dutch SMEs.

This approach seems to be shared with the United Kingdom where responsibilisation is rife. Responsibilisation is the trend whereby the governments impart education on cybersecurity to its citizens but disengage with them if they become victimised. Although, as cyber-attacks evolve, the possible damage to the state through a single infected citizen can soon come to resemble a Hollywood-style blackout with digital infrastructures becoming compromised in life endangering ways.

I also benefitted from visiting the building that houses the NSCR and linking in with my fellow PhD colleague whose research interests are closely aligned to mine albeit with a focus on understanding the offender’s decision-making process in an empirical setup. I experienced it as reassuring to see how we faced similar setbacks in research. It was clear that to succeed in research one must be a highly resilient self-starter prepared to fail a lot and recover frequently and quickly.

In conclusion, I enjoyed my visit to Amsterdam and the NSCR as I had an opportunity to compare the cybercrime situation in the Netherlands with that of the United Kingdom. Since, the problems we share are the same, we should join forces and convey to our respective governments by persuasive research that cybercrime is a worsening international concern. If it is not taken extremely seriously now, then we might not get a second chance in the future.

24 August 2022,

by Dorian Gouzou, Heriot-Watt University

Hi, I’m Dorian, a PhD student at Heriot-Watt University studying machine learning applied to medical imaging. I am (already) finishing my second year and realised that I spent my PhD learning about the different imaging modality I work with, and experimenting basic machine learning on it, which is obviously good because it is my PhD topic, but I haven’t done any deep learning at all, and haven’t kept up with the state of the art of what was my undergrad and master topic.

Thanks to SICSA which granted me funding to go to Prague, I was able to attend the Vision Summer School 3 (VS3) organised by the Czech Technical University, which was the best experience of my PhD so far.

After two flights and one hotel reservations being cancelled, and the first day of the event missed (which started to look like the worst experience of my PhD), I arrived in Prague and to the VS3 on Tuesday. I (re-)learned about deep learning and the best practices on that day, before heading to a climbing session. Because VS3 is a particular summer school where, after the 2 lessons of the day, the speakers are not finished and present us a sport they practice.

Thus, after a day dedicated to robustness, uncertainty and multi-modal learning, I practiced orienteering, then went to learn table tennis after being presented the state-of-the-art in 3D learning and concluded the sports session with badminton after learning about tracking.

The Saturday was dedicated to workshops presented by some of the speakers, who showed us their last research.

Thanks to this summer school, I was able to update my deep learning knowledge, learn about a lot of state-of-the-art applications and meet a lot of PhD students and academics.

19 August 2022,

by Tiwonge Msulira Banda, Robert Gordon University

We all know that an electric motor gets hot when it is running. But do you know that the heat coming from the motor poses a big challenge to engineers that design motors? More generally, have you ever wondered how electric machines, e.g. motors or generators are designed?

The heat that comes from a motor or generator during operation can damage the device if the heat is not properly managed. In trying to model heat transfer in electrical machines, electrical engineers have often relied on numerical techniques, which rely on expert knowledge. However, these techniques are general and computer intensive. Today, data-driven modelling techniques are becoming popular in modelling heat transfer in electrical machines.

Engineers that design electrical machines deal with multiple and often competing objectives. The objectives may include maximising efficiency, reducing cost, and many other desirable operational characteristics. An optimal design therefore involves trade-offs between the different objectives in question. Multi-objective Evolutionary Algorithms, which are a group of algorithms that are inspired by nature, are used in identifying optimal designs. The challenge with the use of these techniques in electric machine design is that the algorithms take long and need significant computing power to evaluate each design to identify the optimal design solutions. Depending on the problem and the number of evaluations required, the process can take weeks or even months to complete.

My PhD research trying to address these two problems: modelling heat transfer in electrical machines and speeding up the optimisation process of electric machine design. The research is part of a collaboration between Robert Gordon University in Aberdeen and Johannes Kepler University in Linz, Austria.

I was awarded a SICSA Saltire Emerging Researcher Scheme grant to undertake a six week research visit at Johannes Kepler University, our collaborating university (from 1 June to 14 July 2022). The purpose of the visit was for me to better understand the domain problem within which my research falls and to build relationships with collaborators. As a computer scientist, we use computing techniques to address real life problems and for one to do a good job, an understanding of the problem is always necessary. This was the basis of the research visit.

At JKU, I was hosted at the Institute of Electrical Machines and Power Electronics (EAL (in German)). EAL is known globally for its research in electrical machines and works closely with the Linz Center of Mechatronics (LCM) a spin-off company from JKU. LCM is one of the world’s leading mechatronics centers and the owners of the SyMSpace, one of the most advanced simulation-optimisation frameworks for mechatronic engineering.
Before the visit, I had made some progress in modelling heat transfer using datasets that we had received from collaborators at JKU. Upon arrival, the collaborators spent time explaining the whole process of how motors operate and the challenges caused by heat. They also explained how the data we earlier received was collected and what each attribute meant. To me this was important because I transitioned from just looking at the data as such, to understanding what each data point meant.

During my stay I worked alongside staff and students from EAL and other visiting engineering students from China and Italy. I held regular meetings with my collaborators who provided feedback to the work I was going and answered questions that I had. I was introduced to more research output on the subject matter as I continued to carry out my research in modelling heat transfer. I also worked on additional modelling problems.

I completed my visit on 14th June 2022 and returned to Aberdeen with a good understanding of electrical machines and the domain problem that my research is addressing. I also know the people at JKU that I will continue to work with for the rest of my PhD programme. In the coming weeks to months, we will continue to work together with collaborators at JKU and produce manuscripts for presentation at relevant conferences.

I believe that my visit strengthened collaboration between RGU and JKU in some way and between Scotland and Austria more broadly.

I would like to thank SICSA for awarding me the grant, collaborators at JKU for hosting me, my supervisors at RGU for their support and admin staff at the National Subsea Centre for making the logistical arrangements. Thank you all.

19 August 2022,

by Adrián Javaloy, University of Saarland, Germany

Hi, I am Adrián, a third-year PhD Student at the University of Saarland, Germany. While Machine Learning has become ubiquitous to our society, often times these systems struggle when presented with data coming from different sources (for example, images and their respective captions). My research therefore focuses on understanding this phenomenon, as well as developing methods that can take full advantage of the diverse information coming from different sources.

Thanks to SICSA’s Saltire Emerging Research Scheme, I went on a two-month research visit to the University of Edinburgh, where I closely worked with Antonio Vergari. Antonio is not only an old friend, but a world-renowned expert in tractable probabilistic models—that is, models that can compute complex probabilistic queries exactly, and in a reasonable amount of time (in stark contrast with current Deep Learning approaches). I had a blast working with Antonio. Early on, we discussed and discarded some ideas we talked about before my visit (that’s how science works!), and started exploring other research directions which we both got excited about, and which we are going to keep working on in the following months.

I feel the most invigorating aspects of my visit were two. First, I felt a step forward in my approach to research by working as a senior researcher who proactively seeks for problems to solve, and how to tackle them alongside other potential collaborators. Second, I was forced to express my ideas clearly: explaining your research topic in thirty seconds is not easy, especially if the person in front of you has never worked on your topic.

More importantly, I got out of my comfort zone. During my visit, I joined the Turing AI Reasoning Workshop, as well as the AI Reasoning reading group, both laying way off my usual research. I also had the opportunity to talk about my current interests at Amos Storkey’s group (followed by a pint with this lovely group). Definitely, my main take-away is the people I met during my visit. Not only I was surrounded by a group of supportive students, but I had the fortune to discuss—and hopefully collaborate in the near future—with Siddharth N. about multi-modal learning, or with Frank Mollica about the acquisition of kinship terms, two outstanding researchers in their respective fields.

Albeit short, this visit has granted me the opportunity to set the foundations for future collaborations, as well as to get to know an incredible environment to which I hope to come back in the future.

16 August 2022,

by Tudor Ferariu, University of Edinburgh

Hello, my name is Tudor Ferariu and I am a first-year PhD Student at the University of Edinburgh School of Informatics. My research topic is automating formal verification for smart contracts for the Cardano blockchain. Proving the absence of bugs is very expensive, and large sums of money have already been lost to various exploits, so being able to formally prove the security of your code is paramount nowadays.

I went on a one-month research visit to Utrecht University, where I worked closely with fellow researcher Jacco Krijnen and their supervisor Wouter Swierstra. Jacco had already visited us in Edinburgh, so it felt only right to return the gesture. Their project focuses on certifying the correctness of compilation for Cardano smart contracts, which in combination with my work, would guarantee security for the entire code pipeline, a much more significant result. Together we coordinated our efforts so that our final products could be compatible with one another. Actively exchanging ideas by being in the same room proved invaluable and we managed to identify the best ways to proceed forward.

During my visit, I was also able to attend a local conference, FP Day, dedicated to functional programming, where I saw a great variety of interesting talks, but most importantly was able to network with the many world-class researchers in attendance. It was during this event that I also got to talk in person with some of my other industrial partners, namely Manuel Chakravarty of IOG and Victor Miraldo of Tweag.Io, with whom I managed to set the foundation for future collaboration.

Frequent meetings with other researchers were great opportunities not only to discuss work but also to socialize. I even managed together with my new friend Jacco to connect some of our respective researcher friends that we knew had similar topics from the two universities. Finishing my relatively short stay with a small presentation on my work, I left with a great deal of positive feedback and ideas on how to move forward. It is my belief that our collaboration has brought mutual benefit both socially and research-wise, and we are still actively working together, albeit online.

None of this would have been possible without SICSA’s Saltire Emerging Research Scheme, for which I am immensely thankful!

16 August 2022,

by Jaspreet Kaur, University of Glasgow

I am Jaspreet kaur, a 2nd year PhD student at the University of Glasgow, James Watt School of Engineering. My research interest is exploiting location and geographical information for adaptive beam forming (a 5G and beyond technology) for transport systems. Currently, I am modelling some buildings and areas on Wireless Insite (a radio planning and ray tracing software) to generate data sets for users around the campus. The dataset is used in beamforming algorithms to steer the beam in the direction of user movement.

Under the SICSA’s Saltire Emerging Research Scheme, I completed a two-month research visit at the Faculty of Informatics, Eötvös Loránd University (ELTE), Budapest, Hungary. I worked closely with Dr Sandor Laki, who is an expert in traffic analytics, IP geolocation, application of software-defined networking and language abstractions for programmable networking elements.

Our project is focused on integrating P4 programming and P4 switch for the Localisation of user equipment. The idea is to use the P4 switch to provide information like the angle/phase from the user equipment to the transmitter/ base station to steer the beam in the direction of or user.

During the visit, we initiated the approach to integrate the P4 switch for the localisation of user equipment. P4 switch holds the reference table consisting of a range of phase values that are calculated by dividing the 500 m area from the transmitter by the grid approach. This reference table is used to match the actual position of user equipment and also this information is feedback to the transmitter to steer the beam/pilot signal towards the user.

I also learned about Hungarian culture, I enjoyed Hungarian food and visited attractions; the best part was trying the Hungarian language with natives. This research collaboration has brought productive research outcomes and exciting network-building with colleagues in ELTE and BME. The next plan is to continue our collaboration in an online manner for further refinements of the current research output and to get the work published.

I would like to thank my kind hosts at ELTE, Budapest and supervisors at UOG, Glasgow, as well as SICSA and everyone involved in the Saltire Emerging Researcher scheme. I wish SICSA will hold the programmes like this in future as well to provide future researchers with a chance to explore the world’s research institutes and excel in their research in collaboration.

 

25 July 2022,

by Jude Haris, University of Glasgow

Hello, I am Jude Haris. I am a second-year PhD at the University of Glasgow in the School of Computing. My research interests are Computer Architecture, FPGAs and DNN accelerators. Thanks to SICSA Research Scholar Funding I was able to attend HiPEAC’s ACACES summer school, 10-16th of July.

The summer school took place in the picturesque town of Fiuggi in central Italy. The summer school covered a variety of topics. We were able to choose 4 out of 12 courses to attend and take part in across the week. Additionally, two keynotes were given, first by Andrea Corbelli, IBM  (Sunday) and then by Maximilian Odendahl, Xilinx (Monday). The summer school also consisted of a careers session (Tuesday) with a panel consisting of academics and industry veterans. We also were given an opportunity to present a poster of our works(Wednesday). During this, I was able to present my current work on DNN acceleration using FPGAs and talk to many like-minded researchers. On the last day (Friday) we were awarded a certificate of attending the summer school and were treated to a final day party.

During the week our courses started at 9 am and lasted until 4:30 pm with coffee breaks and lunch in the middle. The first course of the day I attended was “Post-Modern Computer Architecture: Software Maturity and Dataflow” by Karu Sankaralingam, University of Wisconsin-Madison.

This course was very relevant to my area of work and contained very interesting insights into the problems and some of the solutions that we currently face.

Second I had  “Accelerator-centric System Design” by Mark Silberstein, Technion. The courses looked at accelerators from the perspective of the operating system which was something I had lacked knowledge about.

After lunch, in the third slot of the day, I took “Software-hardware co-designs: The compiler science behind the spark” by Alexandra Jimborean, University of Murcia.  This course was very novel to me and helped me understand some of the basics of compiler problems, as I have not worked in this area, in every lecture I learned something new and interesting. For the final slot of the day, I took “Compilers Challenges for Heterogeneous Architectures” – Henri-Pierre Charles, CEA. This was again similar to the third slot where I learned a lot and gave me a new perspective than just the hardware point of view which is normally the case for me.

Overall the main takeaways for me from the summer school were: that there are a lot of fellow researchers that are working on similar problems and that there is definitely potential for future collaborations; sharing my ideas and learning about other researchers’ work over a lunch break or dinner is completely different from sitting in a zoom call; and also that the best gelato in town was at the top of the hill.

18 July 2022,

by Andrew Murray, University of Strathclyde

Hi, I’m Andrew a 2^nd year PhD student at the University of Strathclyde. My research interests are Optimisation, Planning and Scheduling and Explainable AI and my PhD work involves a combination of all of these technologies.  As an example, consider you have to perform a number of tasks before a deadline (a problem well known to PhD students). The time it will take to complete these tasks is uncertain and driven by external factors (perhaps delays due to Covid), how can you plan and execute these activities such that you optimise the probability that you will complete them.  That’s where my research comes in! In case you want to hear more feel free to give me an email (a.murray@strath.ac.uk). More importantly, I’m here to discuss the Saltire Emerging Researcher exchange that I recently completed at Ericsson AI Research in Stockholm.

Shortly after starting my PhD I was introduced by my supervisor to a number of researchers at Ericsson AI research with whom he had been collaborating for a while on a number of interesting projects. I knew very little about the telecoms industry at the time, however I found the problems they were trying to solve fascinating and highly relevant to my research and was keen to learn more. As a result, I decided to apply for a Saltire Emerging Researcher grant to visit them in their beautiful office in Kista, Stockholm.

Internet service providers do as the title suggests – supply internet services to customers. These services can be made up of a number of components running on nodes within datacentres. With the rise of 5G, services are becoming more and more tailored to the needs of the customers. For example, Massive Machine Type Communications (mMTC) that have arisen due to the influx of household IoT devices (think Alexa) do not require the connection to be lightning quick, whereas things like Ultra Reliable Low Latency Communications (URLLC), the type of information transfer required by self-driving cars, do. Figuring out the optimal way to configure these components in the datacentre, while catering to the varying demands of the customers, is exactly the problem I was working on.

While on the visit I worked alongside researchers from the Machine Reasoning team and implemented a toolchain capable of intelligently solving this problem. The toolchain is due to be validated on a real test case within Ericsson datacentres, with a report summarising the methodology and results to be presented in due course. Sometimes within research, there is a tendency to get lost in the theory and become desensitised to the practical applications of your research. This exchange was a fantastic opportunity to witness first-hand how my research and expertise can be applied to solve important problems within society. I look forward to continuing the collaboration in the future!

While the professional and academic benefits are obvious, I also gained a lot personally from the experience. I spent a lot of time at weekends exploring the beautiful islands around the Stockholm archipelago (some photos included) – an experience I would recommend to anyone thinking of visiting Stockholm.

I would like to thank my kind hosts at Ericsson AI research, as well as SICSA and everyone involved in the Saltire Emerging Researcher scheme. Hopefully this brief snapshot of my trip can inspire future emerging researchers to apply to help forge partnerships and drive innovation in Scotland and abroad. Cheers!