education

Education


Round Corner
Department of Computer and Information Science

Fordypningsprosjekt 2019

Marker valg for å avgrense hvilke oppgaver som skal vises.

Hovedprofil

Datateknologi (506)



Helseinformatikk (18)


Information Systems (67)


Datateknologi (506)







 
Faglærere (32)
































Sorter etter:

Oppgaveforslag (125)

[Learning2Program] Learning Analytics for TDT4100

Læringsanalyse er systematisk innsamling og analyse av data om læringsaktiviteter og sammenhengen med læringsutbytte. Målet er å få mer innsikt i hva som skaper god læring og bruke det til å forbedre våre emner.

Faglærer: Hallvard Trætteberg     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

[Learning2Program] Læringsstøtte vha. læringsanalyse

Læringsanalyse (learning analytics) er teknikker for å samle inn og analysere data om læringsprosessen, så en kan gi bedre læringsstøtte. Vi jobber med å ta dette i bruk i TDT4100 og andre programmeringsfag.

Faglærer: Hallvard Trætteberg     Status: Valgbart     Egnet for: En student     Lenke: plink

3D Facial Reconstruction from Front and Side Images

Biometric recognition from 3D facial images is inherently advantageous compared to 2D facial images, as it does not suffer from pose and illumination variations. However, many existing databases consist of one or more 2D facial  images.

Faglærer: Theoharis Theoharis     Status: Tildelt     Egnet for: En student     Lenke: plink

A cognitive architecture integrating generalization-based and case-based reasoning

A Cognitive Systems wind is blowing in AI today, illustrated by several initiatives in the AI and Machine Learning communities, recent dedicated issues of AI Magazine, IEEE Expert, funding agencies like DARPA, EU, etc. The enormous success of purely data-driven machine learning methods is unquestionable, but it is commonly agreed that these methods alone will not solve many of the challenges that lie ahead, such as broader and more robust problem solvers, learning from a few examples (as we humans are good at), systems that can explain their reasoning and justify its conclusions, and interactive decision support systems. Cognitive system architectures are inspired by human cognition and mental models of problem solving and learning, explicit representations, and symbolic processing of knowledge.

Faglærer: Agnar Aamodt     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

A plausible way to understand simple correlations in Norwegian texts

Sticos are developing a bot called @else (http://else.sticos.no) to help Human Resources (HR) departments be more efficient in their work. A lot of their time is consumed by reoccurring questions from their employees and managers. For years HR software have tried to tackle this problem by using a personnel manual. However, the use of it is sparse. The problem lies in accessibility and the fact that is easier to ask the question directly and get a qualified answer to your problem on the fly.

Faglærer: Rune Sætre     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

AI & eHealth: Case Base Evolution

A interesting problem in personalized decision-support systems for improving an individual patient’s health, is to combine general clinical guidelines with past experiences of that same or similar patients. In the EU project selfBACK, in which we tightly cooperate with the Department of public health and nursing at NTNU, we combine rule-based reasoning with case-based reasoning to capture these two knowledge types. The target problem is Low-Back Pain, and the aim of the project is to develop and thoroughly test a mobile phone app that will give a patient advice on how to improve his/her lower back conditions, in a short or long term. Activity data is continuously captured via a wrist band and used alongside subjective information on pain and functionality. This project will focus oncase-based reasoning and data analytics investigating the evolution of cases over time.

Faglærer: Kerstin Bach     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

AI for Search and Rescue of Lost Persons

Norwegian Search And Rescue (SAR) operations are organised under Hovedredningssentralen (HRS) and are carried out through cooperation between public agencies such as the Police, voluntary organisations such as the Red Cross and specialized search teams, and private companies who have resources appropriate for rescue services.

Hovedredningssentralen together with the IT company InSoft Norge AS wants to develop a decision support system that will gather all available data about a lost person situation and use this to advice a rescue team about the location and state of the person. HRS has a computerized archive system that stores the reports (in Norwegian) for all lost person cases since 2010. However, this system is used only as a storage and for simple access. A lot of experiences, competence and lessons learned are lying in the data storage waiting to be actively used and reused for decision making when a new situation occurs.

Faglærer: Pinar Öztürk     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

AI-based prediction in aquaculture

As fish farming sites move to areas more exposed to harsh wind, wave and current conditions there is a growing need for monitoring and decision support, as well as remote and autonomous operations tied to transport, put out, feeding, sorting, delousing, treatment and slaughtering of the fish. The cost of having to interrupt such operations is substantial.

Faglærer: Kerstin Bach     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

AI-based security attacks

AI and ML are widely used for pattern recognition, intrusion detection, malware classification, and therefore offer promising solutions in cyber-defense. However, recent study has shown that an attacker can leverage the neural processing to stealthily distribute an attack, by concealing malicious files into neural network model. Such attacks can evade existing security detection mechanisms and augment attacking capabilities. The focus of the thesis will include three parts:

- Literature review of malware insertion methods
- Create a vulnerable target system
- Studying how to use AI to create attacks on the target and how to defend against the attacks. For instance, attacks such as DeepLocker [1] that show the malicious use of AI in attack process

[1] Dhilung Kirat, Jiyong Jang, Marc Ph. Stoecklin, “DeepLocker - Concealing Targeted Attacks with AI Locksmithing”, Black Hat USA Conference 2018

Faglærer: Jingyue Li     Status: Valgbart     Egnet for: En student     Lenke: plink

Analytics coming from learner-computer interaction: Multimodal Learning Analytics

Mainstream analytics coming from learning systems (ie, Learning Analytics) often relies on data from digital learning environments such as keystrokes, log files and clicks to interpret the interaction between the learner and the machine. However, the human experience of the world, therefore, human learning needs richer data. In recent years, multimodal data has the potential to help us understand the world around us and interpret the complex learning processes in the field of educational technology researchers to try and build models that can process information from multiple modalities. Although the collection, interpretation, and visualization of multimodal data has been extremely challenging for researchers, Recent technological developments and data science advancements have boosted the growth of non-invasive high-frequency multimodal data collections. For instance, 2D and 3D cameras, wearable sensors, biosensors, infrared imaging, eye-tracking and more IoT devices, all offer the opportunity to enhance the way we collect and analyze learner data to achieve a deeper understanding of the interaction between humans and learning technologies. The proposed project aims to gather published evidence in the area of ​​multimodal learning analytics and conceptualise it into a framework that can lead future research in the area. all offer the opportunity to enhance the way we collect and analyze learner data to achieve a deeper understanding of the interaction between humans and learning technologies. The proposed project aims to gather published evidence in the area of ​​multimodal learning analytics and conceptualise it into a framework that can lead future research in the area. all offer the opportunity to enhance the way we collect and analyze learner data to achieve a deeper understanding of the interaction between humans and learning technologies. The proposed project aims to gather published evidence in the area of ​​multimodal learning analytics and conceptualise it into a framework that can lead future research in the area.

 

Faglærer: Michail Giannakos     Status: Valgbart     Egnet for: En student     Lenke: plink

Artificial Intelligence and Adaptive Stress Testing for Computer Games

Create a system for generating fit objects in a simulated environment. The focus will be on a game environment, and a key takeaway of the project will be how the methods applied may be used in other applications and other simulations.

Faglærer: Ole Jakob Mengshoel     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

Autofix feature of software security vulerability detection IDE plugins


Most of the current software security practices are to test the software using penetration testing at the very late stage of software development. As developers are not well trained to develop secure software, or their software security knowledge is not update, developers introduce software vulnerabilities when writing code.

Faglærer: Jingyue Li     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

Automated Engineering (AI Lab Pitch)

Description

Faglærer: Ole Jakob Mengshoel     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

Automatic detection of pro-eating disorder (pro-ED) social media users

Pro-eating disorder groups (pro-ED) are social media sub-cultures that encourage disordered and dangerous eating behaviours, e.g., Pro-Ana (pro-anorexia), Pro-Mia (pro-bulimia) and Thinspro (Thinspiration, a combination of “thin” and “inspiration”). Automatic detection of users sharing, supporting or following pro-ED content can provide information for understanding and preventing eating disorders, as well as for social media moderation. Data on some such users on Twitter have already been annotated, but to fully apply machine learning algorithms such as deep learning to the problem, more data need to be gathered, tentatively from various sites such as Twitter, Reddit and Tumblr. The thesis work would then experiment with applying various machine learners to this data.

Faglærer: Björn Gambäck     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

Autonomous agents for car simulator with Way AS.

The project is a collaboration with the company Way that has developed a car simlator for drivers training. Several topics related to developing functionality for Way's car simulator are available:

Faglærer: Odd Erik Gundersen     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

Autonomous vehicles (AVs), AI/ML/DL and Computer Vision (CV)

Interested in AVs, AI and CV?

Have a look here

 

 

Faglærer: Frank Lindseth     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

Bankrupcy prediction (AI Lab pitch)

Background: This project/master thesis proposal is a collaboration between the  Norwegian Open AI-Lab and BDO.BDO provides a range of services within the areas of Audit, Accounting, Consulting and Tax/Legal. BDO Norway employs more than 1450 people and has more than 70 offices throughout the country. Our clients include major, global companies to small and medium-sized enterprises Internationally, in most areas of the private and public sectors. BDO is
present in 162 countries and employs more than 74 000 people.

Faglærer: Helge Langseth     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

Bankruptcy prediction - with Trondheim Kommune

 There will be a co-supervisor in Trondheim kommune. The task is to predict and detect the danger of companies for bakruptcy. 

Faglærer: Pinar Öztürk     Status: Valgbart     Egnet for: En student     Lenke: plink

bio-inspirerte metoder

Oppgavene skreddersys til studenters interest.

Faglærer: Pauline Haddow     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

Biometric Identification of Salmon

Salmon have a unique "fingerprint" in the form of dots on their skin and head. This has been proven to be unique by SINTEF in their research:
https://www.sintef.no/prosjekter/identifikasjon-av-lakseindivider-biometri-fase-1/

Faglærer: Theoharis Theoharis     Status: Tildelt     Egnet for: En student     Lenke: plink

Blockchain-based Data Marketplace

The European regulations about privacy, the GDPR, is changing people’s understanding and attitude towards data/information they own and those the others own. The ownership of data/information has always been a concern but with GDPR there will be more awareness as well as more obligations related to data collection and sharing.

Faglærer: Hai Thanh Nguyen     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

Case-based reasoning in machine learning and problem solving (open theme)

This is an open project, for which you are encouraged to come with your own ideas. The only constraint is that is should address case-based reasoning in some way or another. The focus can be on knowledgemodelling, knowledge representation, machine learning, problem solving and/or evaluation. CBR could be in combination with other methods, or as a stand-alone method. Theoretical studies, design-oriented AI, or experimental studies are all possible approaches.

Supervisor: Agnar Aamodt

 

Faglærer: Agnar Aamodt     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

ChatBots - Dialog interfaces - Text / Phone

We have several systems that make it possible to ask natural language queries over Internet, by SMS or by voice over telephone about various tasks, e.g bus routes or telephone information. You can try yourself by calling +47 7352 1290, or checking http://busstuc.idi.ntnu.no

Faglærer: Rune Sætre     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

Classification of species and biomass estimation (AI Lab pitch)

Problem Description: The focus and challenge of this project is to improve the quality of detection, biomass estimation and classification of fish and mammal species using machine learning (ML). For estimating biomass, acoustic sonar systems are preferable as their range far exceeds light-based system like cameras and lasers. For identification of (near range) objects, camera images may be better. In this work a sensor system (with a subsea camera pointed in the same direction as the sonar) able to combine both modalities is explored.

Faglærer: Helge Langseth     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

Code-switching and multi-lingualism in social media

When two individuals who are bi- or multi-lingual in an overlapping set of languages communicate, they tend to switch seemlessly and effortlessly between the languages (codes) they share. Such code-switching is most prominent in spoken language conversations, but also occurs frequently in social media texts that are fairly informal and conversational in nature. The aim of this project is to apply various machine learning methods to such code-switched texts from Twitter, Facebook or Whatsapp, and to, e.g., identify the language of each word or to annotate the texts with part-of-speech tags or utterance boundaries.

Faglærer: Björn Gambäck     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

Combining Deep Learning (DL) and Evolutionary Computation (EC)

Before the recent breakthroughs in DL, many EC researchers believed that the evolution of neural network (NN) weights could defeat backpropagation as a general NN training method. Now, however, the focus has shifted toward hybrid systems in which evolution determines large-scale topological features (such as the number, size and mesoscale connectivity of hidden layers) along with hyperparameters (such as learning and momentum rate, regularization technique, etc.), while backpropagation tunes the weights within that constellation.

Faglærer: Keith Downing     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

Combining physical modelling and Machine Learning for predicting and simulating cyclist performance [Collaboration with Team SKY]

(more information here)

Faglærer: Massimiliano Ruocco     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

Computational Creativity

To be creative, we need to produce something which is new, meaningful and has some sort of value. Computers are able to support humans in creative processes, but to also themselves be creative or to assess if an idea or a product is creative. A master thesis project on computational creativity can investigate any creative field matching the interests and backgrounds of the student or students (language, design, music, art, mathematics, computer programming, etc.), and concentrate on one or several aspects of computational creativity, such as the production, understanding or evaluation of creativity, or on computer systems that support human creativity.

Faglærer: Björn Gambäck     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

Computational Creativity and Art

Creativity can be found in nature and in humans, but also in computers, and entails to produce something which is new. However, just “newness” isn’t a sufficient condition for us to consider an idea to be creative, it also has to have some value and meaning: If a 2 year old draws some lines on a paper, we rarely consider it to be art; while if a grown-up does the same, we interpret it as having some deeper meaning – and if the grown-up signs the paper with a well-known artist name, we attribute both an underlying meaning and a monetary value to it. Creativity is thus something which isn’t only a result of the effort of a producer, but also very much the result of how the result is viewed by the consumer.

Faglærer: Björn Gambäck     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

Computational Lingustic Creativity

Computational linguistic creativity can be aimed at creating systems that either are creative themselves (e.g., generate poetry, write lyrics to music, produce analogies or metaphors; or chatterbots), or try to understand creativity (e.g., identify sarcasm, understand humour or interpret rhymes), or support humans in creative processes (such as PhotoShop in the image domain), or evaluate creativity.

Faglærer: Björn Gambäck     Status: Valgbart     Egnet for: En student     Lenke: plink

Computational Musical Creativity

Computers have been used in music both as support for creativity and as creative agents themselves, and both for the composition of the music scores and for writing lyrics. The first algorithmic composition system appeared already in the 1950s (the Illiac suite, Hiller & Isaacson 1958), and since then rule-based systems, stochastic methods, grammar-based methods, neural networks, and evolutionary methods have all been utilised to compose music, and/or for generating lyrics. A master thesis on the topic could address any of these strands and approaches, depending on the student(s) background and interests.

Faglærer: Björn Gambäck     Status: Valgbart     Egnet for: En student     Lenke: plink

Computer Vision and Deep Learning on Mobile Devices

Several projects related to this topic are available:

Faglærer: Frank Lindseth     Status: Valgbart     Egnet for: En student     Lenke: plink

Cool AI stuff

Bring your own idea. We bring AI.

 

Faglærer: Anders Kofod-Petersen     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

Data miningsfunksjon for AsterixDB

I denne oppgaven skal du utvikle en funksjon (User Defined Function (UDF)) for å lese inn datastrøm inn til et Big Data-håndteringssytem kalt AsterixDB (se https://asterixdb.apache.org). Ideen er at en slik funskjon skal kunne fungere som et verktøy for feks. klassifisering av datastrøm (som feks. Tweets) før dataen blir lagret og håndtert videre i et bigdatasystem for analyse e.l.

Faglærer: Heri Ramampiaro     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

Data/Resource Efficient Deep Learning [2019/2020] [Norwegian OpenAI Lab]

 

Faglærer: Massimiliano Ruocco     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

Decentralized AI

Nowadays, big tech giants, such as Google, Facebook, Amazon, IBM and Microsoft, are dominating the AI market by offering cloud-based AI solutions and APIs. They are collecting user data in one place through free services and systems, analyze it for insights and resell to third parties, such as advertisement companies. This model is centralized AI, which is working fine now. But in the long-run it could lead to monopolization of the AI market. This could also cause unfair pricing, lack of transparency, interoperability, privacy issues and excluding smaller companies from AI innovation. Fortunately, there is the emergence of a decentralized AI market, born at the intersection of blockchain, on-device AI and edge computing/IoT.
In this project, we will investigate the possibility to build a proof-of-concept of a decentralized AI application through blockchain, such as Ethereum. AI agents will train and learn models from their own data. The decentralized AI application to be developed in this project then can combine multiple algorithms/models (developed by different agents) performing different sub-tasks. One of possible applications is Decentralized Autonomous Cars. 1 - 2 students.

Faglærer: Hai Thanh Nguyen     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

Deep Learning for Robotic Grasping of Compliant Objects

This work investigates a generative approach to robotic grasping og soft objects, based on image processing with deep convolutional nets.  The project is done in collaboration with SINTEF Ocean AS and is more fully described here:

Faglærer: Keith Downing     Status: Valgbart     Egnet for: En student     Lenke: plink

Deep Learning to combat with micro-plastic pollution

An estimated 275 million tonnes of plastic waste was produced on a global scale in 2010, with 8 million of those tonnes being introduced to the oceans - about 3% of global annual plastics waste. Once the plastic reaches the oceans, it is broken down into smaller particles(micro-plastic) by being exposed to ultra violet (UV) radiation and mechanical abrasion from wave actions [1].The quantity of plastic waste floating at the ocean surface in 2013 was estimated to be approximately 269,000 tonnes (small macro- to micro-plastic), this estimate does not include plastic in-depth or at the seafloor). The plastic debris can affect the wildlife in multiple ways, such as entanglement- entrapping, encircling, or constricting,ingestion- accidental ingestion or ingestion of prey containing plastic, and interaction- being in contact with plastic debris [1].It is therefore important to be able to detect and collect the plastic waste in nature,before it reaches the oceans. Once plastic waste has reaches a micro-stadium, it is near impossible to collect it and remove it from the water. An analysis on deep sea locations(range from 1176 to 4843m) showed that there was an average abundance of 1 micro-plastic per 25cm3(particle sizes ranging from 75 to 161μm) [2]

Faglærer: Hai Thanh Nguyen     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

Deep reinforcement learning on time-series data

The project aims at investigating whether trading in the energy imbalance market can be done using deep reinforcement learning. 

The research is done in collaboration with TrønderEnergi. It is possible to get a working space at their offices at Lerkendal, sit together with the data science team and work closely with the domain experts.

 

Faglærer: Odd Erik Gundersen     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

Designing Causal Models of Blinding Retinopathy of Prematurity (ROP) with Evolutionary Computation

Retinopathy of Prematurity (ROP) is a serious disease in preterm-born babies that often leads to blindness.  Given several key parameters that have been shown to be helpful in predicting ROP (along with data from several thousand patients), evolutionary computation will be used to design quantitative models of the interactions of these parameters to thereby enhance the predictive power of these parameters.  This work is in cooperation with opthamologists at St. Olav's hospital and is more fully described here:

Faglærer: Keith Downing     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

Deteksjon, klassifisering og clustering av feil og hendelser i kraftnettet gjennom analyse av høyoppløselige måledata og bruk av maskinlæringsmetoder

Beskrivelse: Krav til leveringskvaliteten i kraftsystemet, som omfatter spenningskvalitet (kvaliteten på strømleveransen) og leveringspålitelighet, er spesifisert i lovverket gjennom Forskrift om leveringskvalitet i kraftsystemet (FoL). De siste årene har det vært en betydelig økning i instrumentering og overvåkning av kraftnettet, både gjennom spenningskvalitetsmålere (PQA), phasor measurement units (PMU), smarte målere (AMS) og andre typer sensorer. Kombinasjonen av store datamengder (big data) og statistiske analyser (maskinlæring) muliggjør forbedret feilhåndtering. SINTEF Energi arbeider med å anvende moderne statistiske metoder på høyoppløselige måledata/sensordata i nettet. Målet er å gi tilstrekkelig varsel om at en feilsituasjon er under utvikling så tidlig at den kan unngås, slik at situasjonen kan korrigeres uten at det påvirker forbruker. På veien mot dette målet vil oversikter og analyser av historiske feil være til god hjelp for nettselskap for å kunne gjøre rapportering til myndigheter, undersøke årsaker til feil og mulige løsninger (eksempelvis gjennomføre nødvendige omkoblinger i nett med redundans og utkobling av fleksible laster).

Faglærer: Helge Langseth     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

Digital Doppelgänger

Body tracking without markers, using RGBD cameras, has been around for a while, but is limited by the field of view of the camera, and is often quite skittish.

Faglærer: Theoharis Theoharis     Status: Tildelt     Egnet for: En student     Lenke: plink

DL meets NE: Meta-Learning through incremental evolution of deep learning architectures [2019/2020] [Norwegian OpenAI Lab + OsloMet AILab]

 

Faglærer: Massimiliano Ruocco     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

Dronebasert sanking av sau

Faglærer: Svein-Olaf Hvasshovd     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

Efficient Generalization in Deep Reinforcement Learning (AI Lab Pitch)

In collaboration with Telenor research, this project investigates the design of Deep Reinforcement Learning (DRL) systems that generalize across tasks: they can perform more than one.  For more information, see:

Faglærer: Keith Downing     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

Entity-level sentiment impact analysis in social media texts

A key aspect of sentiment analysis is identifying the target(s) of the opinion, that is, to determine which entities in a text the expressed sentiment relates to. Exploring how textual entities are related to a text’s overall sentiment can yield information on how given entities are portrayed in social media, e.g., on Twitter. This requires the application of sentiment analysis techniques as well as named entity recognition and linking, and the use of heuristic or grammatical features to determine entity relevance and sentiment strength.

Faglærer: Björn Gambäck     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

Evolutionary Algorithms for Language Processing

Natural language processing grapples with an ever-changing and moving target. The focus of study, natural language, is natural because it changes, interacts and evolves in various directions. The bio-inspired computational methods described as evolutionary computation and/or genetic algorithms create computational models that evolve a population of individuals to find a solution to a given problem. This project will investigate how evolutionary computation can been employed in some natural language processing task, ranging from efforts to induce grammars to models of language.

Faglærer: Björn Gambäck     Status: Valgbart     Egnet for: En student     Lenke: plink

Evolutionary Game theory - Development of a platform for studying incentive mechanism for resolving Social dilemmas

A consequence of "defect" (i.e., not cooperating) in game theory is known as "tragedy of commons". The goal of this project is to design a "contest platform" where different incentive mechanism can be tested for their performance in promoting cooperation in public good games (where the optimal behavior of individuals is in contrast with the benefits at the society level). The issue of cooperation (and altruism) has long been studied in economy, artificial intelligence and psychology. A famous example is “tit for tat” (see Axelrod’s Tournament) and studies the role of imitation in cooperative behavior. The project starts with a reproduction of Axelrod’s contest and continues with development of a new mechanism to test. Evolutionary methods will be used in simulation and evaluation of incentive mechanisms. 1-2 students

Faglærer: Pinar Öztürk     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

Evolutionary game theory for Public Good Games – Data sharing

The problem of cooperation has been studied in game theory for long time with a focus on Prisoners dilemma which is a 2-player game. This project topic is about N-player games and focuses on studying and developing mechanism for the emergence of cooperation in societies. There are some known reasons why people “defect” instead of cooperation. There are reasons why they cooperate. This project aims to understand the incentive mechanisms underlying cooperation and to understand their dynamics. This is particulary important for production of public goods.
There are many examples of this overarching problem. Here is one: There is a specific platform for sharing data which may help people in their daily life. This happens only if many people share the data. However, sharing data has a cost (e.g., takes time or may have cognitive load) as well as the benefit from having access to such data. A rational agent would just use the platform for getting the information and let the others do the data-sharing job. The research question is what kind of mechanisms can promote cooperation.

Faglærer: Pinar Öztürk     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

Explainable AI for Maritime Situational Awareness (AI Lab pitch)

Problem Description: The main goal of this research project is to explore the feasibility and usefulness of using computer-graphic simulators for development and validation of explainable AI techniques in maritime scenarios. The work falls within the area of computer vision as an enabling  technology for autonomous ships. Deep networks have shown astonishing performance in image recognition and classification tasks. However, effective training requires huge amounts of images (with corresponding labels), whose acquisition (and preparation) is very expensive. As alternative solution, Kongsberg Digital (in collaboration with NTNU) has explored the use of synthetic computer-generated data in combination with domain adaptation techniques. Results have been promising in aquaculture and maritime domains, where satisfactory performance have been achieved with limited effort on datacollection and preparation. Building on that, Kongsberg Digital wants to investigate the relationships among synthetic-vs-real images in the framework of explainable AI, to assess if computer-graphic simulators and domain adaptation techniques can facilitate and/or improve the design of decision-support systems providing predictions coupled with explanations. 

Faglærer: Helge Langseth     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

Explainable Anomaly detection for wind farm export cable (AI Lab pitch)

Background: Equinor currently operates several offshore wind parks: Dudgeon, Sheringham Shoal and Hywind Scotland, with more to come in the future. All the power produced is carried to shore via one or more export cables. For Dudgeon wind farm, there are two such cables, and they are fitted with a Distributed Temperature System (DTS) where refracted light through a fibre optic cable is used to determine the temperature along the cable. The temperature varies a lot along the cable and with the amount of power going through it. In addition, things like sea temperature, burial depth and seabed density may affect the cable temperature. Hence, it is not trivial to determine whether a given temperature reading is normal or not. An abnormal temperature reading could warn of an impending failure in what is a critical asset for the wind farm, so we would very much like to pick it up as early as possible!

The DTS data, along with data for electrical current, sea temp., burial depth and seabed plow resistance is made available in Equinor’s OMNIA data platform in Microsoft Azure, where it is stored in a Data lake component (HDFS). The combination of scalable storage and scalable compute power available in Azure opens up new possibilities for how to use these data, one of which is to train machine learning models to detect any anomalous temperature readings in the cables.

Faglærer: Helge Langseth     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

Fake news detection

The impact of news articles on society can not be underestimated and as the number of online news is increasing, distinguishing fake news from real news is becoming a challenge for readers. This project focuses on analyzing and tracking news articles from different news sources or social media channels in order to detect fake or misleading news.

Faglærer: Jon Atle Gulla     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

Fish Size and Feature Detection Using GPUs and Cameras with Sonar Data

This project will combine GPU and visual computing using state-of-the-art GPUs and under water cameara data that also records sonar data to try to extract, for instance, the size of fish in a large fish farm to evaluate the health of the fish.

Faglærer: Anne C. Elster     Status: Valgbart     Egnet for: En student     Lenke: plink

Games and playful interactions

This project will aim to obtain an overview of Serious Games, Gamification and playful interaction to identify the purposes of each and how they overlap and complement each other in different situations. This will be done through case studies and relevant examples. Examples of application areas will be in learning and social contexts.

Faglærer: Sobah Abbas Petersen     Status: Valgbart     Egnet for: En student     Lenke: plink

Gender and/or age based author profiling

Until a few years ago, gender-based language studies mainly concentrated on speech. However, social media texts now provide plenty of data for extracting author profiles based on parameters such as gender, age and geolocation, while on the other hand posing new challenges for language analysis due to the often unconventional and abbreviated language used, as well as other characteristica of social media text such as usage of hashtags, emojis, emoticons, code-switching (mixing languages), etc. In addition, many users do not volunteer their actual and true profiles. The theme for this thesis project would thus be investigate automatic (machine learning) methods to either extract and classify author profiles in online texts, or to figure out whether one specific user (or type of user) could have written a given text (i.e., cyber forensics).

Faglærer: Björn Gambäck     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

General Data Protection Regulation (GDPR) and opportunities it brings: analysis of my all-data and generating insights

After enforcement of GDPR (General Data Protection Regulation) (May 2018), all companies and institutions collecting data about individuals are obliged to deliver to people the data they collected about them (e.g., whatever facebook, google, amazon, insurance companies etc collects about me shall deliver the data they collected about me when I asked for it). GDPR will give a chance for people to look into her/his data stored by those companies.
So, what people can do with so much (and rich) data about themselves? It would be super hard for them to analyze, extract insights and even look into the raw data downloaded from Facebook, Google, etc..

Faglærer: Hai Thanh Nguyen     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

Gjenfinning av sau ved hjelp av drone

Faglærer: Svein-Olaf Hvasshovd     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

How might we automate a patient-simulator's verbal communication?

Medical training in the form of nursing scenarios with dolls does currently not involve any communication with the patient simulator (doll). Any information regarding the simulator's state (healthy, leg hurts, arm is in pain etc.) has to be inferred or explained by others.

Faglærer: Rune Sætre     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

Human Activity Recognition from Accelerometer Data in HUNT4

Data captured by body worn sensors provides an excellent opportunity for assessing the physical activity of patients and hence creating behavioral profiles over time. Particularly patients with chronic disease can receive tailored advice on how to increase their activity and hence improve their overall life quality.

Faglærer: Kerstin Bach     Status: Valgbart     Egnet for: En student     Lenke: plink

Human-Machine Symbiosis: How does decision-making change in the age of Big Data Analytics?

The nature of decision-making is changing drastically, both in personal lives and in the business sphere. An increasing amount of decisions are now based on insight that is generated through analytics. Despite this, often individuals are faced with cognitive-overload, conflicted views, or biases that result in non-adoption of insight. This project will be done in collaboration with the Big Data Observatory (https://www.observatory.no) and involve designing a study protocol and collecting and analyzing neurophysiological data (eye-tracking and electroencephalography) from study participants. This will be done with the help of an expert in such tools.

 

Faglærer: Patrick Mikalef     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

ICT and data architecture and Smart cities

 

Faglærer: Sobah Abbas Petersen     Status: Valgbart     Egnet for: En student     Lenke: plink

ICT architecture and IoT/ sensor architecture in buildings

This project will focus on an efficient ICT infrastructure to support smart building and neighbourhoods, by looking at IoT, sensor data and a distributed data architecture. The task is related to the work in Zero Emission Neighbourhood Research Centre http://fmezen.no/about-us/, and will include the following tasks:
- Literature review and state of the art on ICT architectures for distributed data sources.
- Design a distributed ICT architecture for IoT and sensor data from buildings.
- Develop a prototype to use the data and provide some services based on the data; e.g. visualization.
This work may be expanded in a Master project, to include other types of data and data sources, such as open data.

 

Faglærer: Sobah Abbas Petersen     Status: Valgbart     Egnet for: En student     Lenke: plink

IDE plugins for making secure software

 

Faglærer: Jingyue Li     Status: Tildelt     Egnet for: Gruppe     Lenke: plink

Identifying characteristics of persons vulnerable to social media extremism

Extremist groups take to social media since they facilitate cheap, quick, and broad dissemination of messages, and allow for unfettered communication with an audience without the filter or ‘selectivity’ of mainstream news outlets. There have in recent years been substantial efforts to identify members already belonging to extremist organisations and track their Internet activities. The present proposal, however, is primarily aimed at the individuals targeted by the extremists, i.e., persons susceptible to their ideas. The goals would be to profile persons vulnerable to extremism and intercept them before they fully turn to the extremist organisations, and to identify sources of extremism and hate-speech in order to preventively destabilize extremist networks.

Faglærer: Björn Gambäck     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

Identifying Online Hate Speech and Cyber Bullying

During the Spring of 2017, parliamentary committees in Germany and the UK strongly criticised leading social media sites such as Facebook, Twitter and Youtube for failing to take sufficient and quick enough action against hate-speech, with the German government threatening to fine the social networks up to 50 million euros if they continue to fail to remove hateful postings within a week.
With legislation in other countries set to follow, properly identifying hatespeech is a pressing issue, not only for the major players, but also for smaller companies, clubs, and organisations that allow for user-generated content on their sites. Many such sites currently use slow, manual moderation, which mean that abusive posts will be left online for too long without appropriate action being taken or that content will be published with delay (which might be unacceptable to the users, e.g., in online chat rooms).

Faglærer: Björn Gambäck     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

Image analysis for sustainable water and land use

Description of project that this task is a part of: prosjektbanken.forskningsradet.no/#/project/NFR/289725/Sprak=en

 

Faglærer: Odd Erik Gundersen     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

Improved ambulance response intervals with distributed and dynamic placement of resources (AI Lab Pitch)

Proposal for master thesis at Norwegian Open AI-Lab, NTNU, from Norwegian National Advisory Unit for Prehospital Emergency Medicine (NAKOS), and Department of Emergency Medial Communication Centre (EMCC), Division of Prehospital Services, Oslo University Hospital.

Faglærer: Helge Langseth     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

Improved Ambulance Response via Improved Placement of Resources (AI Lab Pitch)

Proposal for master thesis at Norwegian Open AI Lab, NTNU, from Norwegian National Advisory Unit for Prehospital Emergency Medicine (NAKOS), and Department of Emergency Medial Communication Centre (EMCC), Division of Prehospital Services, Oslo University Hospital.

Faglærer: Ole Jakob Mengshoel     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

Industry/SINTEF/NTNU, AI/ML/DL and Computer Vision (CV)

More information about the proposed projects can be found here:

Faglærer: Frank Lindseth     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

Interesting AI project

In this project, the joint interests of advisor and student(s) come together. 

Faglærer: Ole Jakob Mengshoel     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

Interpret Deep learning (DL) algorithm of autonomous vehicles

Many automotive companies leverage deep learning technology to improve object classification and action prediction in their autonomous cars. However, it is difficult to explain how Deep learning (DL) algorithm work, which is crucial for safety-critical decision-making.

Faglærer: Jingyue Li     Status: Valgbart     Egnet for: En student     Lenke: plink

Investigating New GPU Features for Performance (NVIDIA DGX2 and IBM AC922)

Look into how effective are current optimization techniques for GPUs on the newest platforms, including comparing the TeslaV100s on our new DGX2 and IBM AC922 systems with libraries and applications that explore new techniques for the recently announced Turing GPUs from NVIDIA and explore how to use these GPU´s tensor processors for HPC applications.

Faglærer: Anne C. Elster     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

Language Processing for Under-Resourced Languages

A major bottleneck for promoting use of computers and the Internet is that many languages lack tools making it possible for people to access ICT in their own language: the vast majority of the World’s languages are still under-resourced in that they have few or no language processing tools and resources. This is particularly true for sub-Saharan Africa. For example, Ethiopia has some 88 different languages, with even the with four largest ones (Amharic, Tigrinya, Afaan Oromo and Somali) lacking most types of language processing resources. However, the evolution of social media texts has created many new opportunities for building such resources and tools. Thus the HaBiT project at NTNU, U Masarýk (Brno), Addis Ababa U, and U Oslo has extracted large web corpora for those for languages. The master thesis work would investigate using those corpora, e.g., for named entity extraction by creating word embeddings and training deep learners on the data, or for part of speech tagging or language identification.

Faglærer: Björn Gambäck     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

Learning to rank - with diversity

IR-baserte metoder brukes etter hvert på mange andre områder enn klassiske tekst-dokumenter eksempelvis kataloger og webshops, bibliografiske søk, musikk- og filmdatabaser. Informasjonen som indekserer her er ofte beskrivelser av entiteter og inneholder navn, titler, kategorier etc. Her vil tradisjonell IR-ranking basert på tf-idf ol. ofte gi dårlig resultat og brukerne har gjerne mange andre preferanser som er relevante å rangere etter (popularitet, pris, typer osv.).
Optimal kombinasjon av alle rangeringshint bestemmes typisk best med bruk av maskinlæring - en overordnet metode som kalles Learning to Rank.
Et aspekt som kompliserer dette ytterligere er at vi sjeldent egentlig vet hva en bruker er interessert i og derfor også ønsker at det blant første del av resultatsettet skal være en viss diversitet i utvalget så vi øker sjansen for å møte forskjelige informasjonsbehov.

Faglærer: Trond Aalberg     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

LSTM-based models for forecasting intraday power prices (AI Lab Pitch) [2019/2020] [Norwegian OpenAI Lab + Refinitiv]

[Project in collaboration with Norwegian OpenAI Lab]

Faglærer: Massimiliano Ruocco     Status: Tildelt     Egnet for: Gruppe     Lenke: plink

Machine Ethics on Discrimination and Fairness

Ethical issues in Artificial intelligence and machine learning has become a hot topic in the last couple of years. It is a very complicated and broad issue. This project focuses only on one dimension of it: fairness. It is known that human beings make discrimination in employee hiring situation and evaluation of bank loan applications on the basis of gender, ethnic origin, race, etc. Computer systems using machine learning (ML) methods are found to repeat the same kind of discrimination. This project starts with a survey on the state-of the art work on machine discrimination using the existing data sets, and develops metrics for measuring fairness of ML methods. 1-2 students

 

Faglærer: Pinar Öztürk     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

Machine Learning at the Exaflop scale and Beyond (AI Lab Pitch)

Description

Faglærer: Ole Jakob Mengshoel     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

Machine learning for CP Diagnosis

 

Faglærer: Heri Ramampiaro     Status: Tildelt     Egnet for: Gruppe     Lenke: plink

Machine learning to identify consistent biological signals

Contemporary biomedical research generates large quantities of data, where each experiment’s design choice contributes to a potential bias in data. More often than not, the analyses deal with the issue by analysing only a single dataset at time, as the biases are difficult to accurately describe. Ironically, combining datasets with different biases can be used to identify and eliminate the biases, and keep only the genuine biological variation.

Faglærer: Pål Sætrom     Status: Valgbart     Egnet for: En student     Lenke: plink

Machine Learning to improve the Air Quality in Trondheim

Over the last year we have created a dataset with information on air quality data and we started to explore machine learning methods for predicting the air quality for the next 12/24/48 hours as well as visualize the results.

Faglærer: Kerstin Bach     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

Manuell oppfølging av sau på beite

Faglærer: Svein-Olaf Hvasshovd     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

Measuring Prediction Model "Staleness" (AI Lab Pitch)

Description of Problem

Faglærer: Ole Jakob Mengshoel     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

Methods for applying machine learning on sparse datasets

The goal of this project will be two step; the
ultimate goal is to increase the prediction accuracy. Given the amount of data available this will normally be challenging. The second goal will be to develop and test methods for enabling the first goal: The students will be testing new methods that helps state of the art ML methods perform better on small data-sets. Ideas from
applicants are welcome but some starting points could be:
- Discretization (moving from regression to classification as target variable is
originally a continous variable).
- Data expansion via GAN.

 

Faglærer: Bjørn Magnus Mathisen     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

Model Parallelisation of Deep Learning (AI Lab Pitch)

In collaboration with Graphcore and Telenor, this project investigates the efficient partitioning of stochastic gradient-descent modules on Graphcore's state-of-the-art hardware.  For more information, see:

Faglærer: Keith Downing     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

Modeling cell-cycle phase distribution in cell cultures

The cell-cycle is a fundamental molecular process, and disruptions in this process are a hallmark of cancer. When studying eukaryotic cell cycle, the most common approach involves synchronizing a cell culture. In this method, the cells, grown in a medium, are prevented to progress through the cell cycle past a certain phase, usually through use of a chemical agent. The block is then released, and the cells are allowed to progress through the cell cycle.

Faglærer: Pål Sætrom     Status: Valgbart     Egnet for: En student     Lenke: plink

Modelling Smart and Sustainable Cities

ICT plays an important role as enabling technologies in the field of smart and sustainable cities. The Smart cities concept often takes on a limited view of a city and tends to focus on one or few aspects of a city. In this project, we would like to explore the possibilities of modelling a city by looking at it from a holistic way. We would use the ideas of conceptual modelling and enterprise architectures to understand a city. We would also like to explore the city as a complex system and use complex systems modelling approach to simulate how a city evolves. The outcome of the project will be a model of a city. The tasks include:
- Literature review of enterprise architectures for a smart and sustainable city.
- Literature review of how to model a city.
- Design and develop a model of a city as a complex system.
The work could be extended to a Masters project where the model would be further enhanced and evaluated.

 

Faglærer: Sobah Abbas Petersen     Status: Valgbart     Egnet for: En student     Lenke: plink

Native Language Identification

Native Language Identification is the task of identifying the native language of a writer based solely on a sample of their writing in another language. The task is typically framed as a classification problem where the set of native languages is known beforehand. Most work has focused on identifying the native language of writers learning English as a second language. The master thesis work connects to previous work in IDI's AI group and potentially involves participation in a "shared task competition" on Native Language Identification where training and test data is made available by the organisers (such as https://sites.google.com/site/nlisharedtask2013/).

Faglærer: Björn Gambäck     Status: Valgbart     Egnet for: En student     Lenke: plink

Norwegian sentiment analysis at DNB Customer-Insight (AI Lab Pitch)

The Customer-Insight unit at DNB consists of data analysts, data scientists and business analysts who are interested in using machine learning techniques to improve predictions of use-cases such as churn-prevention and up-sales. To do that, the suggest Master's Thesis would investigate sentiment analysis of chat texts in Norwegian (assigning positive, negative or neutral sentiment) and then the usage of this sentiment as input to a supervised classifier to predict churn.

Faglærer: Björn Gambäck     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

Optimizing 4D CT Computation for Performance Through Parallelizations, GPU Computing and AI

This project will be done in collaboration with Prof Dag Breiby´s research group at NTNU Physics, an is part of our joint NFR FRIPRO project on computational microscopy.

Faglærer: Anne C. Elster     Status: Valgbart     Egnet for: En student     Lenke: plink

Personalised recommendations at DNB PULS (AI Lab Pitch)

DNB PULS is DNB's product for corporate banking and includes a rule-based recommendation system which generates advices to a variety of customer types. The suggested Master's Thesis topic would be to explore machine learning alternatives to this rule-based solution, working together with the PULS team.

Faglærer: Björn Gambäck     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

Personalized Recommendations at DNB PULS (AI Lab Pitch)

Background

Faglærer: Ole Jakob Mengshoel     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

Plagiarism Detection

The task of plagiarism detection is to look at a given a document and establish whether it is an original or not. It can be further divided into two sub-tasks: source retrieval and text alignment. The aim of the thesis work would be to solve one or both of them (tentatively depending on whether one or two students worked together). In source retrieval, the task is: given a suspicious document and a web search API, retrieve all plagiarized sources while minimizing retrieval costs. While the task of text alignment is: given a pair of documents, identify all contiguous maximal-length passages of reused text between them.

Faglærer: Björn Gambäck     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

Predict failures in the electrical grid

This task is to use sensor measurements and rule-based alarms to predict failures in the electrical grid to alarm the operators of a real-time operation centre. A task is to analyze the available data to identify how to tackle the problem. Could the rule-based alarms be used as symptoms that indicate problems ahead of time? Is it necessary to find new patterns in the sensor measurements instead of using the rule-based alarms?

Faglærer: Odd Erik Gundersen     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

Prediction of osteochondrosis and skeletal health in pigs using computer tomography (CT) images and machine learning

Norsvin SA have since 2008 been using CT to improve the body composition, health and meat quality of pigs. Each year 3500 pigs get CT-scanned. With the new test station opening in Canada as part of Topigs Norsvin global test system for purebred boars, the number of tested animals increase to almost 10.000 tested animals annually.

Faglærer: John Krogstie     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

Probabilistic reasoning on time-series

This task is done in collaboration with TrønderEnergi.

Faglærer: Odd Erik Gundersen     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

Problems in bioinformatics

The bioinformatics group works on developing and using computational models to predict how changes in gene regulation can control development and cause disease. Towards this end, we develop custom algorithms, statistical simulations, and machine learning-based solutions to analyze and interpret biological data; examples of previous MSc-theses include a genetic programming (GP) approach to predict microRNA target sites, a support vector machine (SVM) approach to identify microRNA genes, and an approach that combines GP and SVMs to identify related proteins.

Faglærer: Pål Sætrom     Status: Valgbart     Egnet for: En student     Lenke: plink

QoS-aware cluster manager using machine learning.

 

Faglærer: Rajiv Nishtala     Status: Valgbart     Egnet for: En student     Lenke: plink

Recognition of Sleep Patterns on Sensor Data Streams (HUNT4)

Data captured by body worn sensors provides an excellent opportunity for assessing the physical activity of patients and hence creating behavioral profiles over time. Particularly patients with chronic disease can receive tailored advice on how to increase their activity and hence improve their overall life quality.
The focus and challenge for this project and master thesis is the selection, implementation and improvement of pattern recognition and data mining techniques to identify sleep patterns from sensor data. The data will be provided by NTNUs medical faculty (DMF), while this thesis should focus on the data analysis. The captured data sets will be streaming data from two acceleration sensors recorded at 100 Hz.
During this work you will build on previous work that does a basic classification of awake/sleep phases and extend the model finding various sleep patterns. Also investigating different sleep stages is a possibility. The implementation will be evaluated in collaboration with DMF, who is also providing background information on the data.
Once the experimental set up is created, different existing algorithms should be evaluated and their strength and weaknesses pointed out. Based on this analysis, a follow-up master thesis can be defined focusing on improving existing algorithms and validated in a real world setting.

Faglærer: Kerstin Bach     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

Recognize objects in junction boxes

Perform object recogition using deep learning on images of junction boxes to understand which components it contains, their names and how they are coupled. The database contains about 15 000 geotagged images. If time allows, it would be interesting to use the geotags and the recognized objects to build a knowledge graph (ontology) of the recognized objects.

Faglærer: Odd Erik Gundersen     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

Reproducibility of AI Research

Reproducing experiments is central in research methodology, also AI research. However, recently the research community in general has uncovered that a large amount of the published research - even research published in the most prestigious journals - is hard or impossible to reproduce, sometimes even for the researchers conducting the research to begin with.

Faglærer: Odd Erik Gundersen     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

Reputation-based incentive mechanisms for emergence of cooperation in Public Good Games

The players in a prisoners dilemma “defect” and receive less benefit compared to what they would get if they had cooperated. However, if they play the game many times (and particularly if they didn’t know how many times) then cooperation may emerge. So, first iterated games have other dynamics than the one-shot games. Second, N-player games are different than 2-player games. Finally, the games are radically different in nature where mechanisms that may emerge cooperation in one play don’t work in another game. These said, this project focuses on “public good games” (e.g., snowdrift game) which involves a contradiction between the rational behavior of individuals and the optimum outcome for the society. Despite this contrast and expectations of non-cooperative behavior according to traditional game theory, in real life cooperation has been observed to emerge. This has been attained to mechanisms such as kinship or close network of people which involve direct or indirect reciprocity. This project will have a focus on the role of reputation in promoting cooperation. Evolutionary game theory methods will be used in simulation and evaluation of designed reputation-based incentive mechanisms. 1-2 students

Faglærer: Pinar Öztürk     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

Reservoir Computing with physical limits

Reservoir Computing (RC) was introduced as a methodology to exploit random recurrent neural networks (RNNs) for computation. Key to RC is a linear readout layer which is trained on the activity of the RNN to produce some desired function. The RNN is referred to as the "reservoir" and remains untrained.

Faglærer: Johannes Høydahl Jensen     Status: Tildelt     Egnet for: En student     Lenke: plink

Risikobasert scoring (COPY)

Oppgaven gis i samarbeid med Sparebank 1 Kredittkort, og den vil veiledes av undertegnede samt en fra Inst. For Matematiske Fag.

Faglærer: Mads Nygård     Status: Tildelt     Egnet for: En student     Lenke: plink

Sau-and Go spill

Faglærer: Svein-Olaf Hvasshovd     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

Scalable blockchain

Blockchain technology today is severely limited by the inability to scale. A throughput of a couple of transactions per second cannot satisfy the real world use cases of public blockchains. Several solutions have been proposed to fix the issue of scalability, and have thus far been divided into two main categories.
The first one focuses on the underlying blockchain technology in order to improve the core protocol to enable scalability without sacrificing the benefits of the current Proof of Work chains. New solutions are being researched in the areas of consensus mechanisms, such as Proof of Stake, Sharding, and even new public ledger solutions that use Directed Acyclic Graphs.
The second category includes solutions that build upon the underlying blockchain technology. Proposed solutions include child-chains, such as Plasma chains and state channels, and solutions that utilize Zero Knowledge Proofs with SNARK or STARK technology.
This specialization project will focus on a literature review of current technologies and proposed solutions. The project must then evaluate and choose one of these technologies, or, if possible, investigate any indications or leads towards new potential solutions.
A potential master thesis can then be used to research the chosen technology and attempt to implement a solution. The aim would be to evaluate the effectiveness of such a solution, and if it solves, or contributes to solving, the issue of scalability in blockchain, without sacrificing the important features of current technology.
References
[1] Dmitry Khovratovich. “State of the Art in Verifiable Computation” (2018). [2] Joseph Poon and Vitalik Buterin. “Plasma: Scalable autonomous smart contracts”. White paper (2017). [3] Zibin Zheng et al. “An overview of blockchain technology: Architecture, consensus, and future trends”. 2017 IEEE International Congress on Big Data (BigData Congress). IEEE. 2017, pp. 557–564.

Faglærer: Jingyue Li     Status: Valgbart     Egnet for: En student     Lenke: plink

Self-Sovereign Identity (SSI) and blockchain

Ever since blockchain came in the light of the greater public, there has been an increasing amount of research done on its use in user authentication. Despite its importance and widespread usage, user authentication is surprisingly still amongst the biggest source of security weaknesses, and as such presents a great amount of potential.
The main advantages of using blockchain in user authentication include its decentralized nature, traceability and integrity.
The potential of this kind of system also lies in the implementation of a true Self-Sovereign Identity (SSI), that may be used across organizations.
This comes with the added advantage of avoiding the single point of failure architecture most user authentication systems use today. Of course, this also comes with its own set of added challenges, including its potential requirement on an initial trustworthy party, and the issue of how to motivate users to validate the blockchain.
Questions still arise on the correct way of implementing such a system, and the exact role the blockchain should have (what should be on the chain: PIIs, authentication requests...).
There is also the question on how to best integrate this with existing systems, and how corporations, who are traditionally more adept of a hierarchical approach, might want to make best use of this.

The aim of the specialization project would be to present an overview of the state of the art, and explore the challenges and advantages,
 

Faglærer: Jingyue Li     Status: Valgbart     Egnet for: En student     Lenke: plink

Sentiment Analysis in Tweets

In recent years, micro-blogging has become prevalent, and the Twitter API allows users to collect a corpus from their micro-blogosphere. The posts, named tweets, are limited to 140 characters, and are often used to express positive or negative emotions to a person or product. In this project, the goal is to use the Twitter corpus to do sentiment analysis, that is, to classify tweets as to whether they express positive or negative opinions, or are neutral/objective. The work could build on previous master theses at NTNU, and potentially aim to participate in a shared task competition on Twitter Sentiment Analysis.

 

Faglærer: Björn Gambäck     Status: Valgbart     Egnet for: En student     Lenke: plink

Sentiment Analysis of Figurative Language in Twitter

The master thesis project is aimed at the automatic classification of tweets containing figurative language, that is, language which intentionally conveys secondary or extended meanings (such as sarcasm, irony and metaphor). Such figurative language creates a significant challenge for sentiment analysis systems, as direct approaches based on words and their lexical semantics often are inadequate in the face of indirect meanings. One goal of the project is to find a set of tweets that are rich in figurative language, another goal is to determine whether the writer of each such tweet has expressed a positive or negative sentiment, and possibly the degree to which this sentiment has been communicated.

Faglærer: Björn Gambäck     Status: Valgbart     Egnet for: En student     Lenke: plink

Societal Sentiment Analysis: Predicting Social Media Personalities, Values and Ethics

Several models can be used to find out how users’ social media networks, behaviour and language are related to their ethical practices and personalities, Such models include Schwartz’ values and ethics model and Goldberg's Big 5 model that defines personality traits such as openness, conscientiousness, extraversion, agreeableness and neuroticism. The thesis project would investigate applying such models to social media text and how the user personalties are reflected by the social networks that they participate in and develop.

Faglærer: Björn Gambäck     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

Støtteverktøy for å lage programmeringsoppgaver for digital eksamen

Transisjonen fra papireksamen til digital eksamen gir potensielle fordeler med mer effektiv organisering av eksamen og mer effektiv sensur. Til gjengjeld kan det være mer tungvint å lage eksamen. For de fleste oppgavetyper vil det gå mye raskere å skrive et spørsmål i en vanlig tekstbehandler enn å legge de inn i Inspera (som NTNU bruker for digital eksamen). Digital eksamen gir også mulighet for nye, interessante oppgavetyper som f.eks. dra-og-slipp - men igjen kan et problem være at disse er forholdsvis tidkrevende å legge inn i systemet.

Faglærer: Guttorm Sindre     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

Supervised Learning of Tennis Activity

In this project, recurrent neural networks will be used to classify time-series sensor data (from a sensor-laden training suit) into strokes and body movements, which then will form the backbone of an automated activity log.  For more details, see:

Faglærer: Keith Downing     Status: Valgbart     Egnet for: En student     Lenke: plink

Systemizing, merging and visualizing data from multiple air emergency operations

Norsk Luftambulanse AS has multiple helicopter bases throughout Norway and Denmark. For each mission a lot of data is generated. This information consists of mission-specific data like response time of the crew and who is partaking as well as patient-related data throughout the journey from pickup to delivery. This data can be structured, visualized and made sense of in a better way.

Faglærer: Rune Sætre     Status: Tildelt     Egnet for: Gruppe     Lenke: plink

Text analytics for detektering av hatefulle ytringer

I denne oppgaven skal du/dere implementere og evaluere tekstbaserte systemer for analyse av hatefulle ytringer i sosiale media.  Se her for mer informasjon.

 

Faglærer: Heri Ramampiaro     Status: Tildelt     Egnet for: Gruppe     Lenke: plink

Trade Routing Problem in Roll-on Roll-off Shipping (AI Lab Pitch)

Description

Faglærer: Ole Jakob Mengshoel     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

TrønderEnergi - Explainable AI and condition-based monitoring

TrønderEnergi has many hydro plants and wind farms that continuously need maintenance. The task is related to investigating how condition-based monitoring can be done in a fashion that helps the users to understand what is failing and how it can be repaired. 

 

Faglærer: Odd Erik Gundersen     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

TrønderEnergi - Transfer Learning

Transfer learning on time-series data in collaboration with the AI department at TrønderEnergi. 

 

Faglærer: Odd Erik Gundersen     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

Uncertainty-aware Deep Learning a.k.a. Deep Bayesian Learning

Description: Deep learning has been reported to improve upon previous state of the art in many traditional machine learning tasks, like image classification, recommender systems, text-to-speech, and so on. Nevertheless, there are still fundamentally problematic issues with these systems, that invite theoretical work on extensions of deep learning towards (traditional) probabilistic reasoning. This is a  topic of some interest, which has lead to nice tools that can be used for implementation/evaluation, like Tensorflow Probability and Pyro (built on top of PyTorch). For typical research trends, see Part III of "The Deep Learning Book" by Goodfellow et al. The selection of interesting research question(s) in this area will depend on the students' interest. 

Faglærer: Helge Langseth     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

Unsupervised contextualisation at DNB (AI Lab Pitch)

DNB has implemented a platform for collecting user activity logs across websites and mobility apps. Specific parts of the logs are manually tagged with rules (business meanings) that are used by the bank to make decisions (e.g., to grant a loan). The process of defining these rules is called "contextualisation" - which is a time consuming, iterative and error prone task. Hence DNB wishes to evaluate various unsupervised (and other) approaches that can help to automate or semi-automate the contextualisation.

Faglærer: Björn Gambäck     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

Unsupervised Learning of Orca Images (AI Lab Pitch)

In this project, unsupervised learning (possibly combined with supervised learning) will be used to cluster images of orca's (a.k.a. killer whales).  This work is in collaboration with "Data for Good" and is described in the following document:

Faglærer: Keith Downing     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

Unsupervised Learning using Athletic Time-Series Data

This project involves a combination of unsupervised clustering techniques with (supervised) deep learning (using recurrent neural networks) to group time-series data generated by sensor-laden athletic training gear.  This is a cooperative effort with US Positronix and is described in slightly more detail here:

Faglærer: Keith Downing     Status: Valgbart     Egnet for: En student     Lenke: plink

Using Evolutionary Algorithms to Investigate Language Evolution

The aim of this project will be to use evolutionary algorithms as a vehicle to investigate the main dynamics in language evolution. Language evolution is a highly multi-disciplinary research field with the main theories involving biological evolution, language learning, and cultural evolution. When trying to understand the origins of languages, we can try to compensate for the lack of empirical evidence by utilizing evolutionary computational methods to create simulations of how language may have evolved over time, e.g., by creating "language games" to simulate communication between agents in a social setting. In general, simulations on language evolution tend to have relatively small and fixed population sizes, something this study could aim to change.

Faglærer: Björn Gambäck     Status: Valgbart     Egnet for: En student     Lenke: plink

Using Machine Learning to aid GPU-based HPC

Explore using ANN and other Machine Learning techniques for auto-tuning of HPC applications.

Details will vary depending on the students interest and background.

See Prof. Elster for details

 

Faglærer: Anne C. Elster     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

Wearable-sensors, Medical Image Computing, Digital Twins (DTs) and the future AI-based ecosystem for self-management of Health

 Interested in Health, Wearables, Medical Imaging, Digital Twins, Cloud Computing, Web and mobile dev. XR, AI/ML/DL, NLP or CV?

Have a look here

 

 

Faglærer: Frank Lindseth     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

XR (VR/AR/MR) for training, simulation and digital twins

Interested in Virtual, Augmented or Mixed Reality (i.e. XR)?

Have a look here.

 

 

Faglærer: Frank Lindseth     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

Your very own project!

The projects I have made available are simply proposals for you to consider. However, the best projects come from motivated students, and if you have your own ideas regarding a project that is related to

Faglærer: Helge Langseth     Status: Valgbart     Egnet for: Gruppe     Lenke: plink
NTNU logo