Page Not Found
Page not found. Your pixels are in another canvas.
Sitemap
A list of all the posts and pages found on the site. For you robots out there is an XML version available for digesting as well.
Pages
About me
About me
Posts
Reproducing SimCLR
Published:
In this blog post I will document my implementation, experiments and findings obtained from the reproduction of the 2020 paper by Chen et al. entitled “A Simple Framework for Contrastive Learning of Visual Representations” or in fewer words SimCLR.
portfolio
Portfolio item number 1
Short description of portfolio item number 1
Portfolio item number 2
Short description of portfolio item number 2
publications
Low Cost FPGA based Implementation of a DRFM System
Published in 2019 IEEE Radar Conference (RadarConf), 2019
Digital Radio Frequency Memory (DRFM) is a technique used to record an incoming Radio Frequency (RF) signal and apply series of time-delays, amplitude scalings and frequency shifts before retransmitting the signal in an effort to deceive a radar system. The design and implementation of a DRFM system on a low cost Field Programmable Gate Array (FPGA) is shown along with an investigation of the performance of the system architecture.
Recommended citation: Michael, Mesarcik (2019). " Low Cost FPGA based Implementation of a DRFM System " IEEE Radar Conference. 1(1). https://ieeexplore.ieee.org/abstract/document/8835754
Deep learning assisted data inspection for radio astronomy
Published in Monthly Notices of the Royal Astronomical Society, Volume 496, Issue 2, 2020
Modern radio telescopes combine thousands of receivers, long-distance networks, large-scale compute hardware, and intricate software. Due to this complexity, failures occur relatively frequently. In this work, we propose novel use of unsupervised deep learning to diagnose system health for modern radio telescopes. The model is a convolutional variational autoencoder (VAE) that enables the projection of the high-dimensional time-frequency data to a low-dimensional prescriptive space. Using this projection, telescope operators are able to visually inspect failures thereby maintaining system health. We have trained and evaluated the performance of the VAE quantitatively in controlled experiments on simulated data from HERA. Moreover, we present a qualitative assessment of the model trained and tested on real LOFAR data. Through the use of a naïve SVM classifier on the projected synthesized data, we show that there is a trade-off between the dimensionality of the projection and the number of compounded features in a given spectrogram. The VAE and SVM combination scores between 65 percent and 90 percent accuracy depending on the number of features in a given input. Finally, we show the prototype system-health-diagnostic web framework that integrates the evaluated model. The system is currently undergoing testing at the ASTRON observatory.
Recommended citation: Michael Mesarcik, Albert-Jan Boonstra, Christiaan Meijer, Walter Jansen, Elena Ranguelova, Rob V van Nieuwpoort, Deep learning assisted data inspection for radio astronomy, Monthly Notices of the Royal Astronomical Society, Volume 496, Issue 2, August 2020, Pages 1517–1529, https://doi.org/10.1093/mnras/staa1412 https://academic.oup.com/mnras/article/496/2/1517/5848205
Improving novelty detection using the reconstructions of nearest neighbours
Published in Array Volume 14, July 2022, 100182, 2022
We show that using nearest neighbours in the latent space of autoencoders (AE) significantly improves performance of semi-supervised novelty detection in both single and multi-class contexts.
Recommended citation: Michael Mesarcik, Elena Ranguelova, Albert-Jan Boonstra, Rob V. van Nieuwpoort, Improving novelty detection using the reconstructions of nearest neighbours, Array, Volume 14, 2022, 100182, ISSN 2590-0056, https://doi.org/10.1016/j.array.2022.100182. https://www.sciencedirect.com/science/article/pii/S2590005622000388
Learning to detect radio frequency interference in radio astronomy without seeing it
Published in Monthly Notices of the Royal Astronomical Society, Volume 516, Issue 4, 2022
Radio Frequency Interference (RFI) corrupts astronomical measurements, thus affecting the performance of radio telescopes. To address this problem, supervised segmentation models have been proposed as candidate solutions to RFI detection. However, the unavailability of large labelled datasets, due to the prohibitive cost of annotating, makes these solutions unusable.
Recommended citation: Michael Mesarcik, Albert-Jan Boonstra, Elena Ranguelova, Rob V van Nieuwpoort, Learning to detect radio frequency interference in radio astronomy without seeing it, Monthly Notices of the Royal Astronomical Society, Volume 516, Issue 4, November 2022, Pages 5367-5378, https://doi.org/10.1093/mnras/stac2503 https://academic.oup.com/mnras/article/516/4/5367/6692884
The ROAD to discovery: Machine-learning-driven anomaly detection in radio astronomy spectrograms
Published in Astronomy & Astrophysics, 680, A74, 2023
As radio telescopes increase in sensitivity and flexibility, so do their complexity and data-rates. For this reason automated system health management approaches are becoming increasingly critical to ensure nominal telescope operations. We propose a new machine learning anomaly detection framework for classifying both commonly occurring anomalies in radio telescopes as well as detecting unknown rare anomalies that the system has potentially not yet seen. To evaluate our method, we present a dataset consisting of 7050 autocorrelation-based spectrograms from the Low Frequency Array (LOFAR) telescope and assign 10 different labels relating to the system-wide anomalies from the perspective of telescope operators. This includes electronic failures, miscalibration, solar storms, network and compute hardware errors among many more. We demonstrate how a novel Self Supervised Learning (SSL) paradigm, that utilises both context prediction and reconstruction losses, is effective in learning normal behaviour of the LOFAR telescope. We present the Radio Observatory Anomaly Detector (ROAD), a framework that combines both SSL-based anomaly detection and a supervised classification, thereby enabling both classification of both commonly occurring anomalies and detection of unseen anomalies. We demonstrate that our system is real-time in the context of the LOFAR data processing pipeline, requiring <1ms to process a single spectrogram. Furthermore, ROAD obtains an anomaly detection F-2 score of 0.92 while maintaining a false positive rate of ~2\%, as well as a mean per-class classification F-2 score 0.89, outperforming other related works.
Recommended citation: Mesarcik, M., Boonstra, A. J., Iacobelli, M., Ranguelova, E., de Laat, C. T. A. M., & van Nieuwpoort, R. V. (2023). The ROAD to discovery: Machine-learning-driven anomaly detection in radio astronomy spectrograms. Astronomy & Astrophysics, 680, A74. https://www.aanda.org/articles/aa/abs/2023/12/aa47182-23/aa47182-23.html
talks
teaching
Analogue Electronics
Undergraduate course, University of Cape Town, Electrical Engineering Department, 2017
Assisted in the teaching of a bachelor’s analogue electronics course. My responsibilities were to create and conduct tutorials and assignments as well as managing the administration of the course.
Programming Multi-core and Many-core Systems
Graduate course, University of Amsterdam, Informatics Institute, 2020
Assisted in the teaching of a Master’s course in high performance computing. I was responsible for planning and managing tutorials on many and multi-core computing frameworks such as CUDA and OpenMP