For more information about the STSM application procedure, see: https://www.cost.eu/uploads/2021/06/Vademecum-V9-28-April-20212-1.pdf

• MÁRCIO FERREIRA.
  Home institution country: Portugal.
  Host institution country: Slovakia.
  Duration (days): 14.
  Title: Machine learning in neutron stars physics.
  Summary: The STSM improved my knowledge on deep neural networks. I became aware of the impact of the deep neural network structure and the data preprocessing on its learning capacity. My future research will benefit from the work developed during the STSM.

• ROBERTO CORIZZO
  Home institution country: Italy.
  Host institution country: fYR Macedonia.
  Duration (days): 31.
  Title: Scalable machine learning methods for gravitational waves.
  Summary: The action has helped me to be involved in a very interesting research problem, and it has allowed me to reach effectively different experts in the fields of physics and machine learning. The STSM gave me the opportunity to perform joint work to achieve some of the goals defined in the COST action. In particular, I was able to exploit my background in machine learning models for predictive modelling and anomaly detection, in combination with the diverse experience of the host in the parallelization of algorithms with Spark and MapReduce. Moreover, I was able to assess the scalability performances of the machine learning models developed, on a distributed cloud infrastructure, thanks to a Microsoft Azure grant held by the host.

• IVAN STAJDUHAR.
  Home institution country: Croatia.
  Host institution country: Italy.
  Duration (days): 14.
  Title: A system for rapid analysis of seismic waveforms.
  Summary: The action has helped me in kickstarting a new research collaboration with a group of researchers in seismology. My time at INGV, Rome, was well spent.

• MÁTÉ FERENC NAGY-EGRI
  Home institution country: Hungary.
  Host institution country: France.
  Duration (days): 5.
  Title: The effect of precision on narrow-band all-sky search algorithm.
  Summary: Offline collaboration can get us 80% of the way. Half of the missing 20% is kicking off the work, discussing either next to a dining table or a white board, while the second half is comes around publication of the results. The STSM has helped a lot in polishing our work to become of production quality.

• MARIA CONCETTA TRINGALI
  Home institution country: Poland.
  Host institution country: Italy.
  Duration (days): 5.
  Title: Training period at GSSI in view of KAGRA underground sound campaign.
  Summary: STSM allowed me to spend a full week dedicated to Newtonian noise cancellation which is one of Cost Action goals. Discussion and brainstorming with other scientists about this kind of noise, which affects performance of GW detectors at low-frequency, were useful and given me new ideas to carried on my research project.

• LUIGIA PETRE.
  Home institution country: Finland.
  Host institution country: Romania.
  Duration (days): 7.
  Title: Machine Learning for Laser Physics.
  Summary: The STSM was about a computer scientist getting a taste of the research/experiments done at the ELI-NP site in Magurele, Bucharest, Romania by physicists in our COST action. ELI-NP stands for Extreme Light Infrastructure for Nuclear Physics I have learnt quite a bit about laser physics during the week spent there We have come at various hypotheses about the analysis we could perform on the possible experimental data. We plan to work together on these ideas, using machine learning algorithms. Meeting the Romanian researchers collaborating on these ideas was a great pleasure. In some ways, the high point of the visit was the talk given by Dr Kip Thorne on gravitational waves. He is such an inspirational speaker!

• LEÏLA HAEGEL
  Home institution country: Spain.
  Host institution country: France.
  Duration (days): 8.
  Title: Improving gravitational wave detection and analysis with machine learning algorithms.
  Summary: The gravitational waves community work in strong collaboration between its members, and notably between the LIGO and Virgo Collaborations. As both detectors rely on similar technology, exchanges concerning their signal analyses are particularly fruitful and lead to a better sensitivity for the multi-detectors network of gravitational waves, enhancing both the detection and analysis of signals. This STSM lead to spend a week dedicated to direct exchanges concerning the improvement of gravitational waves detection and analysis, enabling a faster convergence of the projects with creative brainstorming sessions that would not have occurred in the case of remote communication, therefore serving its purpose of facilitating the development of research between groups.

• LLUÏSA-MARIA MIR
  Home institution country: Spain.
  Host institution country: Italy.
  Duration (days): 5.
  Title: Implementation of GW machine learning code at PIC’s computer center.
  Summary: IFAE became a full member institution of the Virgo Collaboration only in early November 2018. We want to study the coalescence of binary systems and the determination of the mass spectrum and angular momentum of the objects involved. From the beginning, we adopted a deep learning approach in the analysis of the LIGO/Virgo data, since it is particularly suitable for the identification of a well-defined signal pattern over a much larger continuous background. Being really new in this field, personally and as a group, this STSM has helped me to get really acquainted with the technical aspects of the management of large sets of data and in applying deep learning to address classification problems. Even if the STSM lasted only 5 days, the possibility to interact directly with experts in the field and to discuss together with the IFAE students the strategies to follow will certainly result in a much faster progress than having worked alone.

• DARIO JOZINOVIĆ
  Home institution country: Italy.
  Host institution country: Croatia.
  Duration (days): 18.
  Title: Application of CNNs for seismic data analysis.
  Summary: The action and the STSM have helped me in establishing and continuing a collaboration with a scientific group from a different area of science (computer science) from my own (geophysics). It improved my research, that uses methods from computer science, as the computer science researchers provide me with new knowledge and new perspectives on my research. The STSM provided a very good opportunity to make a quick and strong improvement of my research results. It also lead to beginning new research directions within our group and creating new collaborations with other researchers.

• CIPRIAN-OCTAVIAN TRUICA
  Home institution country: Denmark.
  Host institution country: Romania.
  Duration (days): 38.
  Title: Anomaly and Change Point Detection in LIGO Time Series data using Machine Learning
  Summary: The STSM allowed me to perform joint work with the researchers at UPB on the topic of anomaly and change point detection in LIGO time series which is one of the Cost Action goals. After configuring the experimental setup and better designing our usecases, we managed to perform multiple experiments using classical machine learning algorithms as well as neural network architectures and fine-tune our models using hyperparameter tuning. We plan to maintain the collaboration by working on future joint publications to present our findings

• MATTEO-GIOVANNI
  Home institution country: Italy.
  Host institution country: Netherlands.
  Duration (days): 9.
  Title: Planning schedules, data analysis methods and measurement campaigns for seismic arrays in LVC and ET
  Summary: This STSM was an excellent opportunity to talk in person to other people working in my same field to share our knowledge of the topic. Furthermore, I could see and experience the differences that we have in work methods. And, very often, sharing different views of the same problem is the best way to get to something. Actually, It was just necessary to talk together and examine our work in order to understand what could be done, what not and what was not worth spending time on. A much more immediate process than endless e-mail exchanges.

• LUIGIA PETRE
  Home institution country: Finland.
  Host institution country: Ireland.
  Duration (days): 10.
  Title: Towards autoencoders for gravitational waves.
  Summary: I very much appreciate the STSM mechanism that the COST action provides. In a little more than a week, I had the opportunity to approach a topic that I would not have touched in my normal environment – as an academic, one has teaching obligations as well as meetings almost on a continuous basis. I appreciated the free-form discussions, the pondering, the trial coding, and all the concrete steps we were able to achieve. This is exactly why I wanted to join this action, to push me into applying my machine learning background to data in a realistic setting, where we can learn from each other.

• CATALIN LERODEANU
  Home institution country: Romania.
  Host institution country: Ireland.
  Duration (days): 10.
  Title: Developing a Neural Network Autoencoder for the identification of gravitational waves.
  Summary: Personal contact is very important when starting to collaborate with other researchers. STSMs are a great mechanism to establish this initial relationship, find other researchers in the field and kickstart collaborations. For myself, this visit was well spent and we managed to do far mor work than if we would have attempted to collaborate remotely.

• ELENA APOSTOL
  Home institution country: Romania.
  Host institution country: Denmark.
  Duration (days): 18.
  Title: Efficient ML Algorithms for Detecting Glitches and Data Patterns in LIGO Time Series
  Summary: This STSM was of great help to me, as, first of all, it facilitated a close collaboration with the team from another COST member country. Secondly, it allowed us to build an ML- based platform on which, in the future, we will be able to do experiments using other LIGO datasets.

• FILIP MORAWSKI
  Home institution country: Poland.
  Host institution country: Italy.
  Duration (days): 29.
  Title: Machine learning for anomaly detection
  Summary: During the STSM I had a chance to participate in the very interesting scientific problem associated with the searches of anomalies in the gravitational waves data. Beside the improvement of my knowledge about machine learning algorithms suitable in this task, I learned about the configuration of the work environment which will turn extremely helpful in future projects. I developed the Docker and Singularity images which can be used on HPC servers independent from the host system configuration. Overall, I perceive the STSM as an important step in the development of my scientific career and the realization of my PhD research.

• MICHAŁ BEJGER
  Home institution country: Poland.
  Host institution country: France.
  Duration (days): 12.
  Title: Convolutional neural networks for gravitational-wave signal analysis: classification and denoising
  Summary: The mission was focused on studying the real noisy gravitational wave data, containing transient artifacts (”glitches”) and real astrophysical signals. Thanks to the collaborative work during the mission, we now are able to efficiently produce realistic training data for the 1D convolutional neural network classifier, and efficiently detect and denoise astrophysical signals in buried in real detector’s noise using denoising autoencoders and variational autoencoders. These implementation will be used in reliable assignment of the statistical significance of detections and for efficient parameter estimation of the astrophysical signals.