Overview

If you like programming, data analysis, machine learning, tinkering with Internet of Things devices, or mess around with networked computers, you’re in the right place :)

Here below I keep track of available thesis and internships, as well as past ones (to help you get an idea of what to expect based on what your colleagues did in the past). While scrolling through the proposals, keep in mind that most internships may become thesis and most thesis can scale down to an internship, hence if you find something interesting but in the “wrong” category, do not lose hope: contact me and we’ll manage to adapt the proposal to your needs!

Also, proposals are roughly divided in categories, representing the research topic they are meant to deal with: if you find an appealing topic, but not a satisfactory proposal therein, contact me and we’ll discuss your own proposals :) The numbers you see in both categories and proposals of activities therein are not a ranking of importance or a priority score, but merely a reference number to ease communications.

Finally, if you have your own proposal for a thesis or an internship contact me, even if not covering the topics below, and we’ll discuss.

1) Causal reasoning, Bayesian and reinforcement learning in multiagent systems and IoT

Mchine Learning (ML) is now pervasive in our everyday lives (Google translate, Siri speech recognition, etc.) and successfully applied in many application domains where a software (an agent) needs to learn to do something. However, modern software is rarely a monolithic entity that does everything by itself, rather a collection of indipendent services that collaborate to carry out a given task (e.g. Siri is connected to your Calendar, weather service, apps, etc.). Modern software is thus usually a multiagent system where multiple software entities need to coordinate to achieve a given goal.

Various forms of learning are used in multiagent systems (MAS) to let agents individually learn to accomplish a task, or to recognise a given state of the world. Instead, letting agents learn how to interact socially is yet underexplored in scientific literature, but a promising topic to let multiagent systems self-organise to achieve a shared or system goal.

The following activities unfold within this theme.

Thesis

  1. apply either form of learning (reinforcement, structural bayesian, causal) to let agents in a MAS, or devices in a Internet of Things (IoT) deployment, learn from scratch known interaction protocols (e.g. FIPA protocols)
  2. given a pre-built Bayesian network, or causal structural model, let agents interact according to it to accomplish given goals (e.g. in a IoT scenario of choice). Optionally, let agents explain their course of actions by using the model (Bayesian or causal) as argumentation graph
  3. based on open datasets to be found on the web (e.g. assisted living or smart homes domain, this or that), implement causal discovery algorithms from the scientific literature and test them on such data (e.g. this one)

Internships

  1. survey application of causal reasoning (inference + learning) techniques in multiagent systems
  2. survey application of causal reasoning (inference + learning) techniques in IoT
  3. experiment with OpenAI Gym and/or similar for reinforcement learning
  4. experiment with PyAgrum and/or similar for bayesian learning
  5. experiment with MS DoWhy and/or similar for causal modelling, reasoning, (learning?)
  6. experiment with the iCasa smart home simulator to generate a dataset suitable for the ebove learning mechanisms
  7. experiment with Project Malmo for multiagent reinforcement learning
  8. experiment with OpenAI NeuralMMO for massive multiagent reinforcement learning

Concluded

  • Riccardo Santi. “Esperimenti di transfer learning per reti neurali con robot mobili.” Bachelor degree in Management Engineering, 2021/2022
  • Matteo Sigolotto. “Apprendimento autonomo in sistemi IoT mediante Reti Bayesiane.” Master degree in Computer Science Engineering, 2018/2019

2) Argumentation protocols for joint deliberation/action and situation recognition in multiagent systems and IoT

Many daily activities we carry out are assisted by software (e.g. digital assistants like Siri, Alexa, etc.), that (1) exploit artificial intelligence techniques, e.g. Machine Learning (ML), to deliver their services, and (2) work together with many other software entities to accomplish tasks (e.g. Siri is connected to your Calendar, weather service, apps, etc.).

Using ML may result in opaque systems that re not easily comprehended in their decision making by the user (e.g. why is Netflix suggesting that?). Moreover, having systems composed by multiple software pieces raises the problem of establishing how these pieces should coordinate to ahieve the desired goal. Many techniques exist to make ML explainable, that is able to otivate their decisions to humans, and many other exist to define coordination protocols amongst independent software pieces dictating how each piece should behave.

Computational argumentation is a promising research area that could potentially solve both problems at once.

Thesis

  1. develop a software framework for practical argumentation
  2. develop a software framework for playing dialogue games for multiagent negotiation and conflict resolution

Internships

  1. surveying commonsense reasoning state of art
  2. experiment with the Tweety Java library for computational argumentation
  3. experiment with the ArgTuProlog system for computational argumentation

Concluded

  • Dario Ferrari. “Coordinamento di veicoli autonomi basato su tecniche di argimentazione: simulatore ed esperimenti.” Bachelor degree in Management Engineering, 2021/2022.
  • Andrea Bicego. “Progettazione di Servizi IoT basati su Protocolli di Argomentazione.” Master degree in Management Engineering, 2017/2018.

3) Coordination of vehicles for urban traffic management

Being stuck in traffic is not funny for anyone…while self-driving cars could relieve us from the duty of daily driving, cooperative driving is necessary to enable self-driving cars to handle complex situations like crossing intersections, that require the coordination of multiple vehicles to establish a safe crossing order. Urban traffic management is thus an abundant source of coordination problems that future autonomous and connected vehicles will need to tackle to successfully hit the streets. The following activities unfold in this domain.

Thesis

  1. experiment with existing agent-based simulators adopted/adapted in traffic management domain
  2. design and implement an agent-based microscopic traffic simulator

Internships

  1. experiment with the SUMO traffic simulator to implement state of the art intersection crossing strategies
  2. develop a Graphical User Interface (GUI) for a simple simulator developed by graduate student Dario Ferrari

Concluded

  • Dario Ferrari. “Coordinamento di veicoli autonomi basato su tecniche di argimentazione: simulatore ed esperimenti.” Bachelor degree in Management Engineering, 2021/2022.
  • Nicholas Glorio. “Strategie per la gestione delle intersezioni in presenza di veicoli autonomi.” Master degree in Computer Science Engineering, 2020/2021.
  • Marco Gambelli. “Gestione di incroci stradali per veicoli a guida autonoma basata su algoritmi a prenotazione e ad asta.” Master degree in Computer Science Engineering, 2020/2021.
  • Andrea Vitali. “Esperimenti di coordinazione veicolare tramite blockchain.” Master degree in Computer Science Engineering, 2019/2020.
  • Enrico Rossini. “Attraversamento di incroci per auto a guida autonoma: protocolli e simulazione.” Bachelor degree in Management Engineering, 2019/2020.

4) Space-based coordination for smart cities applications

Under construction

Internships

  1. integrate TuSoW and Tile38 for efficient tuple-based, space-aware coordination
  2. experiment/expand Tile38 to deal with arbitrary data (e.g. try geoJSON or implement ad-hoc representation)

5) Beyond AutoML: software engineering applied to machine learning pipelines

Under construction

Thesis

  1. design of a modular machine learning pipeline for predicting health outcomes in COPD/asthma patients

Internships

  1. using PMML/PFA for sharing machine learning models/pipelines across software platforms (e.g. R and Python)
  2. surveying AutoML state of art (e.g. open source libraries)
  3. serving machine learning models over the web

Concluded

  • Gabriele Rinaldi. “Sviluppo di un’applicazione web per la fruizione di modelli predittivi.” Bachelor degree in Management Engineering, 2021/2022.
  • Benedetta Becchi. “Sperimentazione della piattaforma KNIME: creazione di una pipeline di machine learning per il trattamento di dati sanitari.” Bachelor degree in Management Engineering, 2019/2020.

Miscellaneous (e.g. students’ proposals)

Concluded

  • Luca Morini. “Non SprECO: la web app per monitorare il proprio impatto ambientale quotidiano.” Bachelor degree in Management Engineering, 2020/2021.
  • Nicola Romano. “Finanza decentralizzata: dalla tecnologia Blockchain ad una nuova finanza.” Master degree in Management Engineering, 2020/2021.
  • Carla Marangi. “La sicurezza informatica ai tempi del COVID: rischi e misure protettive per lo smart working.” Bachelor degree in Management Engineering, 2019/2020.
  • Alessandra Occhionero. “Analisi della piattaforma AWS IoT Core: gestione di device IoT simulati.” Bachelor degree in Management Engineering, 2019/2020.
  • Luca Bartolini. “Digital Twin per industria 4.0: piattaforme e scenari applicativi.” Bachelor degree in Management Engineering, 2019/2020.
  • Matteo Santacaterina. “Analisi dati GeoJson a supporto di processi clinici: il Mapping DSS in CONNECARE.” Bachelor degree in Management Engineering, 2018/2019.
  • Andrea Canepari. “Geolocalizzazione e stratificazione di pazienti critici su mappa: il caso del Mapping DSS in CONNECARE.” Bachelor degree in Management Engineering, 2018/2019.
  • Jacopo Stefani. “SISTEMA DI SUPPORTO DECISIONALE IN AMBITO CLINICO: VISUALIZZAZIONE DEI PAZIENTI SU MAPPA.” Bachelor degree in Management Engineering, 2017/2018.
  • Andrea Alberini. “Applicazione Java per l’ottimizzazione dell’attività fisica tramite messaggi motivazionali contestualizzati.” Bachelor degree in Management Engineering, 2017/2018.

See also the legacy Unibo page for older thesis.

See also the legacy Unibo page for older projects / internships.