UNIGOU REMOTE - AVAILABLE TOPICS
Topic Overview:
Group recommender systems (RS) are an interesting extension of “single-user” RS. Their aim is to aggregate preferences of multiple users (group members) and recommend to the whole group while delivering recommendations of both high utility and fair w.r.t. all group members. Both in evaluation and generation, group RS may suffer from various biases decreasing their usability. Task of the student (together with the supervisor) will be to identify some of these biases, describe their mechanisms and propose methods to mitigate their effect. The topic follows on the previous work of the supervisor [2,3,4].
Requirements:
Computer science background with focus on machine learning / information retrieval / recommender systems.
Outcomes:
Preliminary materials for a joint publication (evaluation framework, results, research report etc.).
References:
1. Ricci, F. et al (Eds): Recommender Systems Handbook, Springer, 2022.
2. Malecek & Peska: Fairness-preserving group recommendations with user weighting, UMAP 2021.
3. Peska & Malecek: Coupled or Decoupled Evaluation for Group Recommendation Methods? Perspectives@RecSys 2021.
4. Dokoupil & Peska: Robustness Against Polarity Bias in Decoupled Group Recommendations Evaluation, accepted to GMAP@UMAP 2022.
Group recommender systems (RS) are an interesting extension of “single-user” RS. Their aim is to aggregate preferences of multiple users (group members) and recommend to the whole group while delivering recommendations of both high utility and fair w.r.t. all group members. Both in evaluation and generation, group RS may suffer from various biases decreasing their usability. Task of the student (together with the supervisor) will be to identify some of these biases, describe their mechanisms and propose methods to mitigate their effect. The topic follows on the previous work of the supervisor [2,3,4].
Requirements:
Computer science background with focus on machine learning / information retrieval / recommender systems.
Outcomes:
Preliminary materials for a joint publication (evaluation framework, results, research report etc.).
References:
1. Ricci, F. et al (Eds): Recommender Systems Handbook, Springer, 2022.
2. Malecek & Peska: Fairness-preserving group recommendations with user weighting, UMAP 2021.
3. Peska & Malecek: Coupled or Decoupled Evaluation for Group Recommendation Methods? Perspectives@RecSys 2021.
4. Dokoupil & Peska: Robustness Against Polarity Bias in Decoupled Group Recommendations Evaluation, accepted to GMAP@UMAP 2022.
Charles University, Faculty of Mathematics and Physics, Department of Software Engineering
Biases in Group Recommender Systems
Topic Overview:
Traveling Salesman Problem (TSP) represents a typical NP-complete optimisation problem whose computational requirements grows exponentially with increasing the dimension of the problem, i.e. the number of cities that the traveling salesman needs to visit. No general efficient algorithm exists that would guarantee to provide optimal solution in a reasonable time. In most cases the optimal solution for a given problem instance is not known at all. There are various stochastic techniques available (e.g. evolutionary algorithms, ant systems, particle swarm optimisation etc) allowing us to obtain (sub)optimal solutions in a reasonable time which is often acceptable for practical purposes. The goal of this project is to implement (according to a selected existing study) a system for the optimisation of various TSP instances and to try to tune it (by introducing your own ideas) in order to obtain interesting results for TSP instances “as complex as possible”. The results should be evaluated statistically and compared with those presented in the original study.
Task:
1. Familiarise yourself with the TSP, methods and problems of its solution. For example, see the video in [1] or find your own resource.
2. Make a literature review regarding the stochastic optimisation of TSP in recent years (cca since 2015). Also acquire basic principles of stochastic algorithms (preferred areas: Random Search, Simulated Annealing, Ant Colony Optimisation, Particle Swarm Optimisation). Recommended reading is [2].
3. After a discussion with your supervisor choose one article as a basis for your work. Study this article as best as you can.
4. Create an implementation of the system presented in the article, repeat the original experiment(s) and evaluate statistically the obtained results.
5. Propose a modification of the system the aim of which will be to improve the results or to verify your own ideas in the TSP optimisation. Experiment with various modifications, perform your own set(s) of experiments and create a comparative study evaluating the performance of the algorithm(s) and the quality of the resulting TSP solutions.
Hint: The modifications may include the problem representation, the evaluation function or changes in the optimisation algorithm itself. Apply your own ideas and creativity, inspire yourself on youtube, in the literature or during discussions with your supervisor.
Requirements:
Interest in experimental work, creativity and sense of thoroughness and preciseness. Programming skills mainly using existing frameworks and libraries (C/C++, Python or others).
Outcomes:
A presentation or technical report summarizing your method(s) and obtained results.
References:
[1] The Traveling Salesman Problem: When Good Enough Beats Perfect: https://www.youtube.com/watch?v=GiDsjIBOVoA
[2] Anthony Brabazon, Michael O'Neill: Natural Computing Algorithms. Springer Berlin, Heidelberg, DE, 2015, URL: https://link.springer.com/book/10.1007/978-3-662-43631-8
Traveling Salesman Problem (TSP) represents a typical NP-complete optimisation problem whose computational requirements grows exponentially with increasing the dimension of the problem, i.e. the number of cities that the traveling salesman needs to visit. No general efficient algorithm exists that would guarantee to provide optimal solution in a reasonable time. In most cases the optimal solution for a given problem instance is not known at all. There are various stochastic techniques available (e.g. evolutionary algorithms, ant systems, particle swarm optimisation etc) allowing us to obtain (sub)optimal solutions in a reasonable time which is often acceptable for practical purposes. The goal of this project is to implement (according to a selected existing study) a system for the optimisation of various TSP instances and to try to tune it (by introducing your own ideas) in order to obtain interesting results for TSP instances “as complex as possible”. The results should be evaluated statistically and compared with those presented in the original study.
Task:
1. Familiarise yourself with the TSP, methods and problems of its solution. For example, see the video in [1] or find your own resource.
2. Make a literature review regarding the stochastic optimisation of TSP in recent years (cca since 2015). Also acquire basic principles of stochastic algorithms (preferred areas: Random Search, Simulated Annealing, Ant Colony Optimisation, Particle Swarm Optimisation). Recommended reading is [2].
3. After a discussion with your supervisor choose one article as a basis for your work. Study this article as best as you can.
4. Create an implementation of the system presented in the article, repeat the original experiment(s) and evaluate statistically the obtained results.
5. Propose a modification of the system the aim of which will be to improve the results or to verify your own ideas in the TSP optimisation. Experiment with various modifications, perform your own set(s) of experiments and create a comparative study evaluating the performance of the algorithm(s) and the quality of the resulting TSP solutions.
Hint: The modifications may include the problem representation, the evaluation function or changes in the optimisation algorithm itself. Apply your own ideas and creativity, inspire yourself on youtube, in the literature or during discussions with your supervisor.
Requirements:
Interest in experimental work, creativity and sense of thoroughness and preciseness. Programming skills mainly using existing frameworks and libraries (C/C++, Python or others).
Outcomes:
A presentation or technical report summarizing your method(s) and obtained results.
References:
[1] The Traveling Salesman Problem: When Good Enough Beats Perfect: https://www.youtube.com/watch?v=GiDsjIBOVoA
[2] Anthony Brabazon, Michael O'Neill: Natural Computing Algorithms. Springer Berlin, Heidelberg, DE, 2015, URL: https://link.springer.com/book/10.1007/978-3-662-43631-8
Brno University of Technology, Faculty of Information Technology, Department of Computer Systems
Stochastic Optimisation of the Traveling Salesman Problem
Topic Overview:
The topic is “Statistical Model Checking of Approximate Computing Systems” and it is about modelling, simulation and model checking of a special class of systems. Expected results of the internship will represent a solid base for deeper analysis of the so-called approximate computing systems, especially in the areas of finding better cost/quality trade-offs and obtaining data for research publications.
Task:
1) Summarize aspects of the so-called Statistical Model Checking (SMC) and analyse the actual state in the area of modelling and analysis of approximate computing (AC) systems with a special attention paid to their dynamics.
2) Identify SMC means suitable for modelling and analysis of AC systems as well as for evaluation of their attributes and their effects.
3) Model representatives from a selected class of AC systems (such as approximate algorithms or circuits), check their properties by means of SMC and compare them with properties of "accurate" variants of such systems.
4) Evaluate your approach and discuss it from the applicability and validity viewpoints.
Requirements:
Any previous experience with modelling and analysis of systems is welcome. Active interest in the topic, creativity, ability to solve problems independently as well as ability to (self) study are strongly recommended.
Outcomes:
Models of accurate and approximate variants of the selected class of approximate computing systems, experimental results and a short (about 10 pages long) technical report.
References:
According to the promoter’s/supervisor's recommendation – e.g., you can start here:
http://people.cs.aau.dk/~adavid/smc/index.html,
https://ieeexplore.ieee.org/document/9116207.
The topic is “Statistical Model Checking of Approximate Computing Systems” and it is about modelling, simulation and model checking of a special class of systems. Expected results of the internship will represent a solid base for deeper analysis of the so-called approximate computing systems, especially in the areas of finding better cost/quality trade-offs and obtaining data for research publications.
Task:
1) Summarize aspects of the so-called Statistical Model Checking (SMC) and analyse the actual state in the area of modelling and analysis of approximate computing (AC) systems with a special attention paid to their dynamics.
2) Identify SMC means suitable for modelling and analysis of AC systems as well as for evaluation of their attributes and their effects.
3) Model representatives from a selected class of AC systems (such as approximate algorithms or circuits), check their properties by means of SMC and compare them with properties of "accurate" variants of such systems.
4) Evaluate your approach and discuss it from the applicability and validity viewpoints.
Requirements:
Any previous experience with modelling and analysis of systems is welcome. Active interest in the topic, creativity, ability to solve problems independently as well as ability to (self) study are strongly recommended.
Outcomes:
Models of accurate and approximate variants of the selected class of approximate computing systems, experimental results and a short (about 10 pages long) technical report.
References:
According to the promoter’s/supervisor's recommendation – e.g., you can start here:
http://people.cs.aau.dk/~adavid/smc/index.html,
https://ieeexplore.ieee.org/document/9116207.
Brno University of Technology, Faculty of Information Technology, Department of Computer Systems
Statistical Model Checking of Approximate Computing Systems
Topic Overview:
The topic is “Statistical Model Checking of Cellular Automata Systems” and it is about modelling, simulation and model checking of a special class of systems. Expected results of the internship will represent a solid base for deeper analysis of cellular automata systems, design and evaluation of various techniques and obtaining data, e.g., for research publications.
Task:
1) Summarize key terms and concepts related to the so-called Statistical Model Checking (SMC) and analyse the actual state in the area of modelling and analysis of Cellular Automata Systems (CAS).
2) Identify SMC means suitable for modelling and analysis of CASs as well as for evaluation of their attributes and their effects.
3) Model a representative application based on CAS and check its properties by means of SMC.
4) Evaluate model and discuss it from the applicability and validity viewpoints.
Requirements:
Any previous experience with modelling and analysis of systems is welcome. Active interest in the topic, creativity, ability to solve problems independently as well as ability to (self) study are strongly recommended.
Outcomes:
Models of a representative set of cellular automata systems, experimental results and a short (about 10 pages long) technical report.
References:
According to the promoter’s/supervisor's recommendation – e.g., you can start here:
http://people.cs.aau.dk/~adavid/smc/index.html,
https://mathworld.wolfram.com/CellularAutomaton.html,
https://plato.stanford.edu/entries/cellular-automata/.
The topic is “Statistical Model Checking of Cellular Automata Systems” and it is about modelling, simulation and model checking of a special class of systems. Expected results of the internship will represent a solid base for deeper analysis of cellular automata systems, design and evaluation of various techniques and obtaining data, e.g., for research publications.
Task:
1) Summarize key terms and concepts related to the so-called Statistical Model Checking (SMC) and analyse the actual state in the area of modelling and analysis of Cellular Automata Systems (CAS).
2) Identify SMC means suitable for modelling and analysis of CASs as well as for evaluation of their attributes and their effects.
3) Model a representative application based on CAS and check its properties by means of SMC.
4) Evaluate model and discuss it from the applicability and validity viewpoints.
Requirements:
Any previous experience with modelling and analysis of systems is welcome. Active interest in the topic, creativity, ability to solve problems independently as well as ability to (self) study are strongly recommended.
Outcomes:
Models of a representative set of cellular automata systems, experimental results and a short (about 10 pages long) technical report.
References:
According to the promoter’s/supervisor's recommendation – e.g., you can start here:
http://people.cs.aau.dk/~adavid/smc/index.html,
https://mathworld.wolfram.com/CellularAutomaton.html,
https://plato.stanford.edu/entries/cellular-automata/.
Brno University of Technology, Faculty of Information Technology, Department of Computer Systems
Statistical Model Checking of Cellular Automata Systems
Topic Overview:
The topic is “Profiling of Embedded Applications” and it is practically oriented. Expected results of the internship will represent a solid base for deeper analysis of real embedded systems, especially for studying an impact of various development techniques/means and obtaining data, e.g., for validation of existing models or for research publications.
Task:
1) Familiarize yourself with basic terms and principles related to embedded systems as well as with basic development aspects of embedded applications. Summarize your knowledge into a short report.
2) Perform a research in the area of profiling of embedded applications - summarize key terms, concepts and instruments available in software (such as SystemView or FreeMaster) or hardware (such as ARM's DWT unit) for various platforms.
3) Choose an embedded platform (e.g., ARM), operating system (e.g., FreeRTOS) and a profiler (e.g., FreeMaster). Use them to create a simple embedded application and profile it.
4) Based on the agreement with the supervisor, prepare a set of non-trivial embedded applications and prepare a framework for their profiling.
5) Apply the profiling chain of your framework to the set of embedded applications in order to evaluate and present the profiling results to a user.
Requirements:
Any previous experience with the development of embedded systems is welcome. Active interest in the topic, creativity, ability to solve problems independently as well as ability to (self) study are strongly recommended.
Outcomes:
Profiling results for the set of embedded applications and a short (about 10 pages long) technical report.
References:
According to the promoter’s/supervisor's recommendation – e.g., you can start to study materials for FREEMASTER: FreeMASTER Run-Time Debugging Tool (e.g., video tutorials in the TRAINING & SUPPORT section at https://www.nxp.com/design/software/development-software/freemaster-run-time-debugging-tool:FREEMASTER), SEGGER SystemView (e.g., video tutorials in the Video and SystemView Media parts at https://www.segger.com/products/development-tools/systemview/), uC/Probe (http://micrium.com/probe/uC-Probe- UsersManual.pdf), MCUXpresso SWO Trace (https://www.nxp.com/docs/en/training-reference-material/AMF-SOL-ADVANCED-DEBUG-MCUXPRESSO-IDE-PRESENTATION.pdf, https://www.nxp.com/docs/en/quick-reference-guide/MCUXpresso_IDE_SWO_Trace.pdf) and/or materials to similar instruments.
The topic is “Profiling of Embedded Applications” and it is practically oriented. Expected results of the internship will represent a solid base for deeper analysis of real embedded systems, especially for studying an impact of various development techniques/means and obtaining data, e.g., for validation of existing models or for research publications.
Task:
1) Familiarize yourself with basic terms and principles related to embedded systems as well as with basic development aspects of embedded applications. Summarize your knowledge into a short report.
2) Perform a research in the area of profiling of embedded applications - summarize key terms, concepts and instruments available in software (such as SystemView or FreeMaster) or hardware (such as ARM's DWT unit) for various platforms.
3) Choose an embedded platform (e.g., ARM), operating system (e.g., FreeRTOS) and a profiler (e.g., FreeMaster). Use them to create a simple embedded application and profile it.
4) Based on the agreement with the supervisor, prepare a set of non-trivial embedded applications and prepare a framework for their profiling.
5) Apply the profiling chain of your framework to the set of embedded applications in order to evaluate and present the profiling results to a user.
Requirements:
Any previous experience with the development of embedded systems is welcome. Active interest in the topic, creativity, ability to solve problems independently as well as ability to (self) study are strongly recommended.
Outcomes:
Profiling results for the set of embedded applications and a short (about 10 pages long) technical report.
References:
According to the promoter’s/supervisor's recommendation – e.g., you can start to study materials for FREEMASTER: FreeMASTER Run-Time Debugging Tool (e.g., video tutorials in the TRAINING & SUPPORT section at https://www.nxp.com/design/software/development-software/freemaster-run-time-debugging-tool:FREEMASTER), SEGGER SystemView (e.g., video tutorials in the Video and SystemView Media parts at https://www.segger.com/products/development-tools/systemview/), uC/Probe (http://micrium.com/probe/uC-Probe- UsersManual.pdf), MCUXpresso SWO Trace (https://www.nxp.com/docs/en/training-reference-material/AMF-SOL-ADVANCED-DEBUG-MCUXPRESSO-IDE-PRESENTATION.pdf, https://www.nxp.com/docs/en/quick-reference-guide/MCUXpresso_IDE_SWO_Trace.pdf) and/or materials to similar instruments.
Brno University of Technology, Faculty of Information Technology, Department of Computer Systems
Profiling of Embedded Applications
Topic Overview:
The topic is “Characterization of faults in computer-based systems” – it is research oriented. Expected result of the internship is a survey of published data about faults (e.g., fault occurrence times, probability density functions, cummulative distribution functions, mean time to failure, mean time between failures) regarding selected subset of computer-based systems, e.g., their components such as communication interfaces/busses, memories, logic. The survey will represent a solid base for the consequent validation of existing research models.
Requirements:
At least, basic knowledge about computer-based systems (i.e., about their components, structure, operating principle etc.) is required. Any previous experience with doing a survey is welcome; active interest in the topic, as well as ability to (self) study are strongly recommended.
Outcomes:
A 15-25 pages long survey of characteristics of faults/errors regarding computer-based systems.
References:
According to the promoter’s/supervisor's recommendation – e.g., you can start to study materials accessible via https://ieeexplore.ieee.org/xpl/RecentIssue.jsp?punumber=7298.
The topic is “Characterization of faults in computer-based systems” – it is research oriented. Expected result of the internship is a survey of published data about faults (e.g., fault occurrence times, probability density functions, cummulative distribution functions, mean time to failure, mean time between failures) regarding selected subset of computer-based systems, e.g., their components such as communication interfaces/busses, memories, logic. The survey will represent a solid base for the consequent validation of existing research models.
Requirements:
At least, basic knowledge about computer-based systems (i.e., about their components, structure, operating principle etc.) is required. Any previous experience with doing a survey is welcome; active interest in the topic, as well as ability to (self) study are strongly recommended.
Outcomes:
A 15-25 pages long survey of characteristics of faults/errors regarding computer-based systems.
References:
According to the promoter’s/supervisor's recommendation – e.g., you can start to study materials accessible via https://ieeexplore.ieee.org/xpl/RecentIssue.jsp?punumber=7298.
Brno University of Technology, Faculty of Information Technology, Department of Computer Systems
Characterization of Faults in Computer-Based Systems
Topic Overview:
The topic is is about modelling, simulation and model checking of a special class of systems. Expected results of the internship will represent a solid base for better schedulability analysis of real-time systems at the task level.
Tasks:
1) Do a research in the area of modelling and analyzing properties of real-time systems; especially, focus on schedulability analysis of real-time tasks.
2) Identify sources of uncertainty with regard to real-time systems and tasks. Make a survey of methods and tools for real-time task schedulability analysis as well as of approaches to the schedulability analysis problem under uncertainty.
3) Summarize key terms and concepts of the Statistical Model Checking (SMC) technique. Identify SMC means suitable for modelling sets of real-time tasks as well as for evaluating their schedulability under uncertainty; do a research in this area.
4) Propose the flow of the real-time task schedulability analysis process based on SMC. Discuss real-time task sets and uncertainty scenarios for checking the applicability of the process and evaluating it. Present your approach to modeling real-time tasks under uncertainty.
5) Create models of sufficiently representative sets of real-time tasks in order to check their schedulability by means of SMC in various uncertainty conditions.
6) Evaluate your approach and discuss it critically from the applicability, validity and scalability viewpoints.
Requirements:
Any previous experience with modelling and analysis of systems is welcome. Active interest in the topic, creativity, ability to solve problems independently as well as ability to (self) study are strongly recommended.
Outcomes:
Models of representative systems, i.e., real-time task sets and non/preemptive task schedulers, experimental results and a short (about 10 pages long) technical report.
References:
According to the promoter’s/supervisor's recommendation – e.g., you can start here:
http://people.cs.aau.dk/~adavid/smc/index.html,
https://doi.org/10.1007/978-3-642-16561-0_21.
The topic is is about modelling, simulation and model checking of a special class of systems. Expected results of the internship will represent a solid base for better schedulability analysis of real-time systems at the task level.
Tasks:
1) Do a research in the area of modelling and analyzing properties of real-time systems; especially, focus on schedulability analysis of real-time tasks.
2) Identify sources of uncertainty with regard to real-time systems and tasks. Make a survey of methods and tools for real-time task schedulability analysis as well as of approaches to the schedulability analysis problem under uncertainty.
3) Summarize key terms and concepts of the Statistical Model Checking (SMC) technique. Identify SMC means suitable for modelling sets of real-time tasks as well as for evaluating their schedulability under uncertainty; do a research in this area.
4) Propose the flow of the real-time task schedulability analysis process based on SMC. Discuss real-time task sets and uncertainty scenarios for checking the applicability of the process and evaluating it. Present your approach to modeling real-time tasks under uncertainty.
5) Create models of sufficiently representative sets of real-time tasks in order to check their schedulability by means of SMC in various uncertainty conditions.
6) Evaluate your approach and discuss it critically from the applicability, validity and scalability viewpoints.
Requirements:
Any previous experience with modelling and analysis of systems is welcome. Active interest in the topic, creativity, ability to solve problems independently as well as ability to (self) study are strongly recommended.
Outcomes:
Models of representative systems, i.e., real-time task sets and non/preemptive task schedulers, experimental results and a short (about 10 pages long) technical report.
References:
According to the promoter’s/supervisor's recommendation – e.g., you can start here:
http://people.cs.aau.dk/~adavid/smc/index.html,
https://doi.org/10.1007/978-3-642-16561-0_21.
Brno University of Technology, Faculty of Information Technology, Department of Computer Systems
Schedulability Analysis of Real-Time Tasks under Uncertainty
