Distributed Systems

Research Internships

Metastable Failure Prediction Using ML Techniques

Supervisor: Kawsar Haghshenas, Heerko Groefsema.
Status: Taken (unavailable).
Date: 04/05/2023.
Metastable failures have recently been introduced as a new class of failures in distributed systems [1-2]. As defined in [1], metastable failures occur in open systems that lack control over the source of load, where a trigger causes the system to enter the state of permanent overhead with a low throughput that persists even when the trigger is no longer active. Unlike many other failures attributed to hardware malfunctions or software bugs, the root cause of metastable failures is not a specific hardware failure or a software bug. Consequently, metastable failures are hard to predict and incur the substantial human engineering efforts required due to the inherent difficulty in achieving automated recovery. In [2], three example applications on which metastable failures are experimentally reproduced are presented. In this project, we aim to explore using various ML techniques and the data set created by the experiments in [2], for metastable failures prediction.

  1. Bronson, Nathan, et al. "Metastable Failures in Distributed Systems," Proceedings of the Workshop on Hot Topics in Operating Systems, 2021.
  2. Huang, Lexiang, et al. "Metastable Failures in the Wild," 16th USENIX Symposium on Operating Systems Design and Implementation (OSDI 22), 2022.

Graph Neural Network for Leakage Detection in Water Distribution Networks

Supervisor: Dilek Dustegor, Andres Tello, Huy Truong.
Status: Taken (unavailable).
Date: 12/04/2023.
The project is related to leak detection and localization in water network. This is mainly an implementation project. A recent study [1] has proposed a leakage detection and localization method in water distribution network, combining data-driven techniques (more specifically, graph neural-network, GNN) and model-based logic. The project consists in implementing / reproducing the proposed methodology. However, the GNN-based reconstruction algorithm will be replaced with our own model (Andres and Huy’s model). The corresponding code will be provided for this module.

In the ideal case, the student is expected to:
1- study the paper [1] and implement the proposed method,
2- develop a working understanding of the Pytorch Geometric library (which is used for the implementation of Graph Neural Networks)
3- replace one module with an existing GNN model,
4- perform the necessary optimization of hyperparameters
5- design a set of experiments to validate the leak detection approach
6- run the experiments to collect the performances
7- compile a comparative table
8- discuss the obtained results
However, if time becomes an issue, the focus will be on above listed tasks 1-3. The main deliverables will be a written report describing your work and the documented code. An oral presentation will conclude this internship.
References:

  1. Graph-Based Learning for Leak Detection and Localisation in Water Distribution Networks.
  2. Pytorch Geometric library.

Graph Forward-Forward on Water Distribution Networks

Supervisor: Huy Truong.
Status: Taken (unavailable).
Date: 12/04/2023.
Have you ever trained a Deep Learning(DL) model without backpropagation? Forward-Forward (FF) algorithm is an alternative approach to answering the question. In particular, FF updates the model weights using positive and negative examples through two forward passes. In this research, we apply FF to train a Graph Neural Network (GNN) to monitor the Water Distribution Networks. The model should recover hidden signals from a limited number of existing sensors. As a deliverable, results from FF on GNN can be used for further analysis and comparison to the classical training on diverse water network topologies. One extended question is whether the FF training approach can provide plausible continuous learning for deep models, especially in changing scenarios.

References:

  1. Hinton, Geoffrey. "The forward-forward algorithm: Some preliminary investigations." *arXiv preprint arXiv:2212.13345* (2022).
  2. Paliotta, Daniele, et al. "Graph Neural Networks Go Forward-Forward." *arXiv preprint arXiv:2302.05282* (2023).
  3. Hajgató, Gergely, Bálint Gyires-Tóth, and György Paál. "Reconstructing nodal pressures in water distribution systems with graph neural networks." *arXiv preprint arXiv:2104.13619* (2021).

Research Internship on ECiDA Project

Contact: Mostafa Hadadian.
Status: available.
Date: 15/01/2024.
ECiDA aims to narrow the gap between the data scientists who experiment with data and try new models, and the production environment wherein data science models should eventually run and exhibit consistent behavior. Our goal is to provide a solution that can also be applied to existing data processing platforms.

The infrastructure of ECiDA revolves around containerized components, in which each computational component is responsible for a single step in the data science pipeline. For this, we rely on technologies such as Docker, Kubernetes, and Kafka to containerize, orchestrate, and enable communication between the computational components.

The topics that you can apply include but are not limited to:

  • Monitoring systems
  • Web development
  • Container Orchestration
  • CI/CD Pipelines
  • Network Service Mesh

Automated reasoning using automated planning

Supervisor: Heerko Groefsema.
Status: available.
Date: 26/10/2021.
Verification entails proving or disproving the correctness of a system model with respect to its specification. Such specifications are often expressed using formal methods of mathematics such as temporal logics. To obtain information on successor states in system models, it is possible to rewrite temporal logic expressions using semantic equivalences and expansion laws. Automated planning is an artificial intelligence technique that aims to find an optimal set of actions which together accomplish a predetermined goal. The question for the student then is, can we use automated planning to obtain the possible expanded logic expressions and can we obtain the optimal expanded expression?


Make a difference in Energy Transition with Machine Learning

Contact: Frank Blaauw..
Status: available.
Location: eWEning star.
Date: 01/06/2021.
eWEning star is a “fresh from the oven” Start-Up, which is currently developing a discovery tool that serves stakeholders in the renewable energy sector with relevant scientific information regarding renewable energy. Currently people in this sector use key-word based search queries in order to find scientific papers and reports, but with eWEning star’s concept, these papers are smartly categorized, saving users a lot of time and nerves. By making the search process more efficient we can make the energy transition towards renewables faster! Currently we have around 900 documents that are manually categorized in three different ways: (i) perspective, (ii) position in value chain, and (iii) geographical location. Combined, we have created 15 categories. Depending on the length of your internship, it is possible to work on these all, or choose one out of the three options. While this manual approach is feasible for a small number of papers, it does not scale well. Our aim is to apply Machine Learning to improve this process. We expect that machine learning can provide us with a fast solution for categorizing already published papers according to eWEning star concept. You are given the freedom to design, develop and test a process which leads to the automated categorization. You have a background in Data Science and/or computer science, and you have natural curiosity for solving issues. You aren’t afraid to ask questions if you seem to “hit the wall”, but are capable of working independently. Some entrepreneurial mentality is a benefit as eWEning star is a Start-Up. Good communication skills are needed towards non-technical founder.

Researchable in-company internship

Contact: Frank Blaauw.
Status: available.
Location: Researchable B.V.
Date: 01/06/2021.
Researchable B.V. is a small startup located in Groningen. They aim to improve science by developing software in the early phases of research projects (e.g. developing software to collect data, or automate other parts of research) and at the final phase of research projects (i.e., the valorisation of research). During this internship, the student will be part of the Researchable team, and work on various projects that they are currently running. Their office is located on Zernike.


Have your own project suggestions?

We are available to supervise projects on various aspects of distributed systems, in particular involving

  • Service-Oriented and Cloud Computing
  • Pervasive Computing and Smart Environments
  • Network Centric Real-time Analytics
  • Energy Distribution Infrastructures
  • Adaptive Communication Middleware

If you have an idea of a specific project or would like to work generally in a specific area, please let us know about it and we can then narrow the project down.

Please feel free to contact us to discuss specific topics and options.