education

Education


Round Corner
Department of Computer and Information Science

Fordypningsprosjekt 2017

Marker valg for å avgrense hvilke oppgaver som skal vises.

Hovedprofil

Datateknologi (539)



Helseinformatikk (18)


Information Systems (65)


Datateknologi (539)







 
Faglærere (30)






























Sorter etter:

Oppgaveforslag (131)

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

Accelerating Simulations with the Maxeler FPGA-based DFE (Data Flow Engine) and/or Tegra TX2/Volta

This project is for students that like to experiment with programming on new and challenging FPGA-based hardware.

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

Analysing blogs with a view to dialects, age-groups and sociolects

The Stockholm-based company Gavagai is working on sentiment analysis (extracting opinions) from blog texts. Their systems could also be used for analysis blogs according to dialects, and building tools which would include geo-positioning of the blogs, map mash-ups, statistics, and significance estimates – based on the blog data as such and on the social networks created by and around the blogs. Important applications for this may include identifying whether a person in a social network really belongs to the group he/she claims to belong to, for example to warn against possible grooming attempts (when grown-ups pretend to be children in order to befriend a child, e.g., with the aim to later sexually exploit it).

Medveileder: Jussi Karlgren, Gavagai

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

Applying Machine Learning in Aquaculture

This project will try to apply different machine learning methods to help support operations and decision making in the aquaculture industry and on aquaculture sites.

More specifically we would apply different ML methods on data provided by the aquaculture industry (via SFI Exposed or other similar projects) to predict operational parameters, e.g.: fish welfare, fish feeding, fish growth, HMS danger levels on operational sites, predictive maintainance of sites.. 
These predictions could be based on e.g. camera feeds from the aquaculture cage, movement sensors, close proximity bouy sensors.
 

 

Faglærer: Bjørn Magnus Mathisen     Status: Tildelt     Egnet for: En student     Lenke: plink

Automated cross-document summarization of news storylines

With the advance of digital media, the need for resources to deal with dynamic and complex text data coming from newsstreams is increasingly growing. Particularly, manually tracking down news storylines in newstreams can be a time consuming activity.

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

Automated Security Board Game for Agile Development

Gamification is seen to be useful in Agile development and has been applied to security. Examples of popular security games are protection poker and elevation of privilege. Field results however show that security expertise and experience are needed for useful deliverables when these games are used. Besides, they are manual and requires more than one person to play.

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

Automatic document timestamping

Document timestamping (or document dating) is the process of determining the creation date of a document based on its contents. This can be done, e.g., with methods based on language models [1] or term burstiness [2]. In this project, the aim is to 1) perform a survey of previous approaches to document dating, 2) develop new approaches for this task (for example using "facts" from knowledge bases), and 3) compare the new approaches with previous approaches.

Prerequisites: good programming skills.

[1] http://www.idi.ntnu.no/~noervaag/papers/ECDL2008.pdf
[2] http://dl.acm.org/citation.cfm?doid=2600428.2609495

Faglærer: Kjetil Nørvåg     Status: Tildelt     Egnet for: Gruppe     Lenke: plink

Automatic Fault Surfaces Extraction from Seismic Volume

This project is associated with computational sciences activity group under the collaboration framework between Schlumberger, the largest oilfield services company in the world, and IPT (Petroleum Engineering & Applied Geophysics) and IDI (Computer and Information Science) at NTNU. The aim is to automatically extract fault surfaces from seismic volume with the least amount of human intervention.
Interpreting faults in 3D seismic data is still one of the most time-consuming and tedious aspects of seismic interpretation. Automatic Fault Extraction (AFE) is a process designed to automatically interpret fault surfaces from 3D seismic volume. The AFE workflow is comprised of the following three main steps explicitly or implicitly:
A. Fault attributes: select an appropriate fault-sensitive attribute to highlight the fault location
B. Fault likelihood: transform the attribute volume into a fault likelihood/confidence volume
C. Fault surfaces: generate a localized surface mainly in form of polygons, from the confidence volume
Over the last decade, a multitude of AFEs have been developed, including kinds of fault attributes (e.g. chaos, curvature), various fault likelihood (ant tracking, hough transform, skeletonization, applying geological constrain) and different fault surfaces (e.g. in format of fault sticks, polygon meshes). Although a few of them has been implemented in the commercial software but it is not satisfying the end users.
Exploring new methods to further improve the performance and reduce the human intervention will accelerate the interpretation workflow by orders of magnitude. Thus it continuously attracts lots of researchers from kinds of disciplines and try to treat the AFE from different viewpoints. With our background in computer science, it would be possible to at least investigate two sub-disciplines:
1) Computer vision: A possible solution in mind would be ‘active contour model’, also called ‘snakes’. It is a framework in computer vision for delineating an object outline from a possibly noisy 2D image. The snakes model is popular in computer vision, and snakes are greatly used in applications like object tracking, shape recognition, segmentation, edge detection and stereo matching.
2) Machine learning (Optional): Neural Network has been used for fault extraction in OpendTect. Other choice from machine learning would be support vector machines (SVMs). It is a supervised learning models with associated learning algorithms that analyze data used for classification and regression analysis. In the context of fault extraction, the classification is fault or not fault. It is optional depending on the progress in item 1.
The project is ambitious and could lead to a publication. It will be implemented as a Plugin for Petrel, which is a Schlumberger owned E&P software platform that provides an integrated solution from exploration to production. Novel results will also be compared against A3Mark our seismic attributes benchmark for quantitative results.

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

Automatic Fault Surfaces Extraction from Seismic Volume using Machine Learning

This project is associated with computational sciences activity group under the collaboration framework between Schlumberger, the largest oilfield services company in the world, and IPT (Petroleum Engineering & Applied Geophysics) and IDI (Computer and Information Science) at NTNU. The aim is to automatically extract fault surfaces from seismic volume with human intervention as less as possible.
Interpreting faults in 3D seismic data is still one of the most time-consuming and tedious aspects of seismic interpretation. Automatic Fault Extraction (AFE) is a process designed to automatically interpret fault surfaces from 3D seismic volume. The AFE workflow is comprised of the following three main steps explicitly or implicitly:
A. Fault attributes: select an appropriate fault-sensitive attribute to highlight the fault location
B. Fault likelihood: transform the attribute volume into a fault likelihood/confidence volume
C. Fault surfaces: generate a localized surface mainly in form of polygons, from the confidence volume
Over the last decade, a multitude of AFEs have been developed, including kinds of fault attributes (e.g. chaos, curvature), various fault likelihood (ant tracking, hough transform, skeletonization, applying geological constrain) and different fault surfaces (e.g. in format of fault sticks, polygon meshes). Although a few of them has been implemented in the commercial software but it is not satisfying the end users.
Exploring new methods to further improve the performance and reduce the human intervention accelerates the interpretation workflow by orders of magnitude. Thus it continuously attracts lots of researchers from kinds of disciplines and try to treat the AFE from different viewpoints. With our background in computer science, it would be possible to at least investigate machine learning for fault extraction. A choice from machine learning would be deep neural network (DNN). It uses a cascade of many layers of nonlinear processing units for feature extraction and transformation. Each successive layer uses the output from the previous layer as input. The algorithms may be supervised or unsupervised and applications include pattern analysis (unsupervised) and classification (supervised). In the context of fault extraction, the classification is fault or not fault.
The project is ambitious and could lead to a publication. It will be implemented as a Plugin for Petrel, which is a Schlumberger owned E&P software platform that provides an integrated solution from exploration to production.
Requirements
[Necessary] Familiar with Linux, Good programming skills in Python, TDT4195 (Visual Computing Fundamentals)
[Preferably] TDT4230 (Graphics & Visualization), TDT4265 (Computer Vision), Machine Learning.
Dataset to be used: Will be provided by Schlumberger
Advisors: Liyuan Xing (NTNU), Victor Aarre (Schlumberger), Theo Theoharis (NTNU).
If interested, please contact Liyuan Xing, preferably by 10.5.2017. liyuan.xing@ntnu.no

 

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

Automatisk klassifisering av transaksjoner

SpareBank 1 SMN ønsker å tilby sine kunder god oversikt over sitt privatforbruk. Per i dag ligger det et filter i nettbanken som er manuelt oppbygd. Dette vil kreve vedlikehold for å være oppdatert når det kommer en transaksjon mot f.eks. en ny butikk vi ikke allerede har klassifisert. Men gjennom å for eksempel koble butikknavnet mot Brønnøysundregistrene, kan man raskt klassifisere transaksjonen ut fra hvilken næringskode butikken har. Dette vil fungere mot varetransaksjoner, men ikke mot overføringer til f.eks. kundens bankkonto i annen bank, eller når det betales husleie til en privatperson. For å få en god automatikk må det trolig brukes flere ulike datakilder og metoder for klassifisering.

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

Autonomous Perception and Deep Learning (DL), inc. game-based data augmentation

Several projects related to this topic are available:

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

Autotagging og anbefalingsmotor for nyhetssaker

SpareBank 1 SMN ønsker å kunne tilby sine næringslivskunder relevante nyheter. En måte å gjøre dette på kan være å innhente dagens nyheter fra relevante nyhetskilder, autotagge innholdet, og tilby riktig type nyhetsinnhold til hver kunde. Dette kan styres enten basert på kjennetegn ved kunden (nystartet bedrift, sektor, størrelse, geografi) og\eller på kundens egne oppgitte preferanser for innhold (ønsker å lese om ny teknologi, børs, etc).

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

Avoiding biases in building training datasets for supervised algorithms in NLP

RQ: How can we build a training dataset that is unbiased in the context of words used (vocabulary), sentiment(positive and negative connotation) and topic for the purpose of training a natural language processing engine to extract keywords, or do topic modelling? (Optimizing for TFIDF and LDA)

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

Behavior analysis for anticipating market trends

The goal of the project is to investigate how can AI/Machine Learning can be utilized in behavior analysis in order to anticipate market trends to automatically pre-optimize market activities.

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

Beslutningsstøttesystem basert på diverse sensorinput / IoT

Telenor-NTNU AI-Lab.

Faglærer: Jo Skjermo     Status: Valgbart     Egnet for: En student     Lenke: plink

Big data for social good and social innovation - Co-supervisor Ilias O. Pappas

The purpose of this student project (master thesis) is to contribute to the body of knowledge with innovative practices, guidelines, policies, which will help entrepreneurs leverage big data to better design social innovation.

Faglærer: Maria Letizia Jaccheri     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

BIO-INSPIRED: Classifying Time-Series Data via Long Short-Term Memory

Long Short-Term Memory (LSTM) neural networks, invented in 1997 (by Hochreiter and Schmidhuber) have enjoyed a renaissance due to the advent of Deep Learning.  Since 2013, LSTM has been successfully employed for tasks such as speech recognition, machine translation and image captioning.  This project will explore the use of LSTM in detecting motion patterns in time-series sensory data.

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

Bio-INSPIRED: Deep Dreaming

This project will begin by exploring Google's recent system, DeepDream, based on Tensorflow, which uses a procedure known as algorithmic pareidolia (simulated hallucination) to produce creative visual patterns based on previous visual experiences.  The essence of DeepDream will then be implemented in the student's own Tensorflow model and encapsulated within a stylish user inferface that will enable users to a) easily enter images, b) train a deep network to recognize similar images, and c) produce hallucination images corresponding to any neuron (in any layer) that the user chooses.  Other features may also be included, limited only by the student's own creativity.

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

BIO-INSPIRED: Deep Reinforcement Learning for Dynamic Channel Allocation

In cellular telephone networks, each new caller must be allocated a channel (within the provider's frequency band) that does not interfere with those allocated in nearby areas.  This quickly becomes a serious problem when many callers congregate in particular areas, such as entertainment venues.

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

BIO-INSPIRED: Deep Reinforcement Learning for Gripper Vector Estimation

SINTEF Ocean is leading a research project, called iProcess (www.iprocessproject.com ) aiming to develop novel concepts for flexible robot based automation in the food processing industry. An interesting area of research in WP3 (http://iprocessproject.com/wp-3-flexible-processing-automation/) consists of a vision-guided and machine-learning robot that can grasp/manipulate compliant food objects. Given the 3D images from a RGB-D camera, the aim is to develop a grasping concept that finds a suitable grasping pose for the gripper tool. In principle, there are different learning strategies to establish and enable kinematic grasping such as Learning from Demonstration (LfD) and Reinforcement Learning.
In this project assignment, the student will implement Deep Reinforcement Learning for 6DOF pose estimation based on RGB-D images. A dataset of RGB-D images is already collected but the student may collect additional data for training. In reinforcement learning, the goal is to maximize the overall reward for successful grasping, and therefore, in contrast to LfD there is a metric function that enables improvement of learned behaviour i.e. grasping of the compliant food object.

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

BIO-INSPIRED: Understanding and Visualizing Deep Convolutional Networks

SINTEF Ocean is leading a research project – Intelligent- aiming to develop novel concepts for salmon behaviour prediction in net cages. The concepts and methods developed in Intelligent inspired by state-of-the-art BigData concepts and exploits Deep Learning to enable improved context awareness and intelligence for monitoring and control of operations in aquaculture industry. An interesting area of research is Computer Vision and the use of Convolutional Neural Networks(CNNs). These networks learn to represent images through learning increasingly advanced filters as the network grows deeper. Recently, video classification tasks have also seen improvements through the use of CNNs and this has also been explored in the Intelligent project.
In this project assignment the student will investigate what Convolutional Neural Networks are actually learning through training on video data. The student will have access to a large dataset of underwater videos of salmon in net cages. The student will analyse and investigate the filters in several trained deep CNNs to investigate what the networks are learning from the videos and to vizualize the outputs of each layer in the networks. One example deep learning architecture the student will analyze in detail is the 3D-CNN architecture. This architecture is able to capture temporal information from videos by using stacks of video frames as inputs. This enables the network to learn more useful features from videos.
The assignment is as following:
- Using the dataset of existing videos and using existing trained models, explore and visualize what the networks have learned and how this impacts the model performance.
- Train new models, using the results from the analysis, to improve model performance on prediction accuracy.
A state-of-the art workstation with the necessary software to work with videos will be made available to the student.
Prerequisites:
- Excellent programming skills in Python
- Knowledge and interest in deep learning, machine learning and computer vision concepts.
Co-supervisors from SINTEF Ocean:
Dr. Ekrem Misimi – Senior Scientist (ekrem.misimi@sintef.no) and Håkon Måløy

Faglærer: Keith Downing     Status: Tildelt     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: En student     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

Computer Vision, Deep Learning & Industry (EFI, Trollhetta)

Several projects proposed by industry related to computer vision and deep learning are available (more information can be found at the end):

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

Context-Sensitive Recommender Systems for Anglers

Recommender systems are today common on online news sites and shopping sites.  The systems take into account the user's preferences and/or similarities with other users, but are otherwise indifferent to the user's context.

Faglærer: Jon Atle Gulla     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

Cyborg brain-body-computation (kunn for Ton)

Prosjekt for Cherdsak Mangmee

 

Faglærer: Gunnar Tufte     Status: Tildelt     Egnet for: En student     Lenke: plink

Data Analytics for HUNT: Recognition of Physical Activity on Sensor Data Streams

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: Tildelt     Egnet for: Gruppe     Lenke: plink

Data Integration and recommender systems at TNS Gallup, Oslo

TNS Gallup advises clients on specific growth strategies around new market entry, innovation, brand switching and customer strategies, based on long-established expertise and market-leading solutions. With a presence in over 80 countries, TNS has more conversations with the world’s consumers than anyone else and understands individual human behaviours and attitudes across every cultural, economic and political region of the world. In Norway, TNS is the largest provider of market research and insights, with approximately one third share of the market. We work with a wide variety of large and small companies with both a local and international presence. TNS is part of Kantar, the data investment management division of WPP and one of the world's largest insight, information and consultancy groups.

Faglærer: Jon Atle Gulla     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: En student     Lenke: plink

Data miningsmetode for monitorere kvalitet og sikkerhet ved utredning av sykdom

Målsetting: Utvikle og validere metoder for bruk av data mining/maskinlæring til å monitorere kvalitet og sikkerhet ved utredning av sykdommen Polymyalgia revmatika (PMR). PMR er en betennelsestilstand i sene- og muskelfester som arter seg som generelle smerter og stivhet i kroppen og som behandles med kortison.

Potensiale for effekt: Behandling som gir resultater av verdi for pasienter forutsetter korrekt diagnose. Prosjektet skal utvikle og validere metoder for monitorering av diagnostiske feil som gir verdifull tilbakemelding til behandlende helsepersonell og kan benyttes som et grunnlag for lærende prosessforbedringer.

Kunnskapsstatus: Omlag 1 av 20 pasienter som utredes ender opp med forsinket eller feil diagnose. Feil diagnose kan føre til at pasienter utsettes for behandling på feil indikasjon. Konsekvensene er betydelig helsetap, tap av tillit og store erstatningssaker. Kvalitet og sikkerhet i diagnostiske prosesser er et lite utforsket område: Det er ingen automatikk i at kliniske avdelinger får tilbakemeldinger på eget diagnostisk arbeid. Det er derfor et behov for å utvikle egnede metoder for å overvåke forekomsten av diagnostiske feil samt kartlegge hvilke faktorer som påvirker forekomsten, herunder det å innhente pasienterfaringer. Det vil bli brukt kvantitative data (registerdata for sammenstilling av data på tvers av avdelings- og institusjons-grenser).

Oppgaven vil bli utført i samarbeid med Arild Faxvog (lege) og Pieter Toussaint.

 

Faglærer: Heri Ramampiaro     Status: Valgbart     Egnet for: En student     Lenke: plink

Decision support for emergency call centers

SmartHelp is a mobile app that enables anyone calling to the emergency call centers using a smart phone to share relevant data, so that the emergency call centers get to understand where the emergency is, what the problem is and who is calling.

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

Decision support for managing pesticide application in agriculture

A challenge in agriculture is to manage the pesticide application in way that the application is most effective, cause least side effects and being most cost effective. Therefore a farmer has to individually answer the following questions:

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

Decision support in patient-centered health care

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 (http://www.selfback.eu), in which we tightly cooperate with the Department of public health and general practice 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 either on the case-based method, the rule-based method, or their combination. A concrete project specification will be worked out that takes the candidates’ interests, as well as the needs of the project ,into account.

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

Decision support in patient-centered health care

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 (http://www.selfback.eu), in which we tightly cooperate with the Department of public health and general practice 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 either on the case-based method, the rule-based method, or their combination. A concrete project specification will be worked out that takes the candidates’ interests, as well as the needs of the project ,into account.

The project is suitable for 1-2 people working in a group, but can also be defined for one person.

Supervisor: Kerstin Bach and Paul Jarle Mork, ISM.

 

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

Deep Learning and Reinforcement Learning for Recommender Systems

[Project in collaboration with Telenor-NTNU AILab]

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

Deep Learning for NLP - Question/Answering, Chatbot and Text Understanding

[Project in collaboration with Telenor-NTNU AILab]

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

Deep Learning for Robo-Journalism

[Project in collaboration with Telenor-NTNU AILab]

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

Deep Reinforcement Learning

[Project in collaboration with Telenor-NTNU AILab]

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

Define your own project in Visual Computing

This one is open to students who have an interest and an idea for a project in Visual Computing.

Students will be chosen based on their performance in the Visual Computing courses and the suitability of their proposal.

Send email by 10. May 2017 to theotheo@idi..ntnu.no

 

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

Defining roles in transaction networks using deep learning

[Project in collaboration with Telenor-NTNU AILab]

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

Developing data structures for industrial assets

Industrial companies typically have many physical assets such as equipment parts, sensors, turbines, plants etc. Assets also have relationships, e.g a sensor is part of a compressor which is part of a drilling platform. The goal of the asset database is to provide a single entry point for all data related to an industrial asset, such as documents (PDFs), ERP data, maintenance logs, 3D models, and timeseries sensor data. The user should also be able to answer queries such as what are all the parts of a given machine, list all my centrifugal pumps, and what equipment is situated inside a 3D bounding box.

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

Discovering Interpretable Temporal Communities for User Behavior Prediction in Online News

The study of users’ behaviour prediction has become more and more important in news recommendation area. But in real news networks, the context of user actions is constantly changing and co-evolving, and consequently, traditional methods such as matrix factorization and collaborative filtering can be ineffective due to the volatility of individual’s behaviors and sparsity of individual’s interaction with news articles.

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

DNB sin Chatbot (MVP) på facebook messenger.

DNB sin Chatbot (MVP) på facebook messenger og fremtidige chatbots i nett- og mobilbank vil ha 2 moduser. Chatten vil først behandles av en robot, men kunden vil også få valget om å chatte videre med en rådgiver.
I denne oppgaven er det interessant å finne ut:
1. Hvor gode er våre modeller på å finne intent og gi et riktig svar til kunden?
2. I hvilken grad aksepterer kunder svaret som gis av roboten?

Faglærer: Rune Sætre     Status: Tildelt     Egnet for: En student     Lenke: plink

Dronebasert sanking av sau

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

Evaluation of methods and metrics for quality assessment of a topic modelling and keyword extraction algorithms

RQ: What is the best way to evaluate topic modelling and keyword extraction algorithms in terms of quality?

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

Evaluering og verifisering av semantiske data

Mye data gjøres tilgjengelig som semantisk web data - spesielt fra bibliotek, arkiv og museer. Dette er data som er skapt over mange tiår med forskjellige standarder og formater.  Når data legges ut som linked open data, er det gjerne basert på transformasjon til modeller med eksplisitt semantikk. Utfordringen er at store mengder av det som publiseres er av dårlig kvalitet og bare utgjør støy som ikke har noen gjenbruksverdi. 

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

Evolution of altruism in Internet of Things (IoT)

Co-supervisor: Hai Thanh Nguyen  (Telenor)

Faglærer: Pinar Öztürk     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

Explanation-aware army builder for Warmachine and Hordes

Warmachine and hordes are strategic games using miniature figures, where each of two players control one army. One of the challenges is constructing a "good" army, based on the rules of the games. Choices made as to the nature of the troops, their equipment, and so forth is typically a function of opponents and player style.

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

Exploring interfaces for learning at the workplace

This work is a part of a industry project focused on designing technologies for situated learning at the workplace. The project aims to explore different technologies for learning such as Serious Games and Simulation-based solutions. This project will focus on the different types of interfaces that could be relevant in this context. The student will be required to try one or more of the technological solutions that are currently available and evaluate its use in the appropriate setting. The evaluations must be conducted on a theoretical and practical basis.
In addition, the student will be required to obtain an overview of patterns and trends in technology adoption and use in recent years and an overview of new and emerging technologies that are relevant for future workplaces.

 


(Suitable for both single and groups of students)

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

Extracting structured data about incidents from text maintenance logs

Industrial companies typically have large amounts of sensor data from operating equipment such as turbines, pumps and compressors. The sensor data may represent pressure, vibrations, heat etc. over time. In order to use machine learning to predict equipment failures before they occur, it is necessary to obtain training data about past incidents from historical maintenance logs. Due to the large volume of such data from different companies, we seek to automate parts of this process by training an NLP algorithm on a subset of the logs and let it label the remaining.

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

Failover in MySQL

Automatic failover in MySQL. Suits students interested in distributed systems.

Guidance by Oracle.

 

Faglærer: Svein Erik Bratsberg     Status: Valgbart     Egnet for: En student     Lenke: plink

Four Campuses One (Virtual/Mixed) Reality

Several projects related to the "Four Campuses One Reality" project (that has funding to create an infrastructure of VR labs across the campuses at Gløshaugen, Dragvoll, Øya, Gjøvik and Ålesund) are available:

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

Fra hendelsesdata til aggregert analyseinput

En av de mer tidkrevende oppgavene vi møter på når vi skal lage prediktive modeller er å finne en optimal aggregering av data på hendelsesnivå (klikk i nettbank, transaksjon på konto etc.). For at slike data skal kunne benyttes i tradisjonelle prediktive analyser, må de aggregeres for hver kunde for å kunne fungere som en forklaringsvariabel. Et eksempel på et slikt utfall vi ønsker å predikere kan være kundeavgang, mislighold eller produktkjøp.

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

From Gradients to Surfaces using Machine Learning

Further work on the project described at Uppsala (in collaboration with Victor Aarre (Schlumberger):

Faglærer: Theoharis Theoharis     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

Gjenfinning av sau ved hjelp av drone

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

GPU implementation of registration of Point Clouds from a RGB-D camera mounted in a robot arm

SINTEF Ocean is leading a research project, called iProcess (www.iprocessproject.com ) aiming to develop novel concepts for flexible robot based automation in the food processing industry. An interesting area of research in WP3 (http://iprocessproject.com/wp-3-flexible-processing-automation/) consists on the vision guided and machine learning based robot manipulation of compliant food objects. Given the 3D images and point clouds from a RGB-D camera mounted on a robot arm, the aim is to develop a grasping concept that finds a suitable grasping pose for the gripper tool.

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

High Performance Graph Processing on Modern Hardware

Graph processing exists in many practical problems in computer science. Most graphs from real applications, such as transportation routes and social network analysis, are large scale and complex structured. This nature makes high performance graph processing on modern heterogeneous parallel devices more interesting.

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

Hybrid Heuristics for Multiobjective Integrative Optimization

Industrial optimization problems are mostly complex and involve the solution of different interrelated optimization problems. Given the dependency between decision-variables of different optimization problems, it is necessary to combine them into a global integrative optimization problem. This project considers a formulation by combining k optimization problems into a multi-objective integrative optimization (MIO) problem with k objectives. In practical, no single solution is optimal, but there is a set of alternatives representing the trade-off resulting from the integration of underlying optimization problems. The main challenge of this project is how to design effective heuristics for MIO problems. The interaction between decision-variables raises non-trivial aspects on the design of heuristics. As proof-of-concept, the heuristics are analyzed in two real-life manufacturing and logistic MIO problems.

Faglærer: Kazi Shah Nawaz Ripon     Status: Tildelt     Egnet for: En student     Lenke: plink

Hybrid machines

At the CARD group biological/digital hybrid machines is a resent approach to unconventional computing machines. The morphogenetic engineering initiative and the NTNU Cyborg project are project including such hybrid computers.

Faglærer: Gunnar Tufte     Status: Valgbart     Egnet for: En student     Lenke: plink

IDE plugins for making secure software

 

Faglærer: Jingyue Li     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

INTELLIGENT ROBOTICS: Deep-Learning Robots

Robots are commonplace in society; they perform all sorts of jobs. However, most are hard-wired to perform a few fixed tasks and have little or no ability to adapt to changing or unforseen circumstances. Artificial Intelligence (AI) enters the picture when robots must be capable of autonomous behavior in unpredictable environments. In this project, the student must utilize AI machine-learning techniques to produce a robot with adaptive capabilities.

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

INTELLIGENT ROBOTICS: Neural Episodic Control

In a recent paper ("Neural Episodic Control", Pritzel et. al., March 2017), the engineers in Google's DeepMind laboratory (London) present a memory model that combines neural networks with a sophisticated evaluation table (known as a differentiable neural dictionary).  The resulting system achieves considerable speedups on the Atari-game task that highlighted Google's ground-breaking, Deep Learning research in 2015.  The two most appealing features of neural episodic control are: 1) a combination of a long- and shorter-term memory into the same system (akin to the interplay between neocortex and hippocampus in the mammalian brain), and 2) a completely differentiable model, meaning that the full power of automatic gradient processing in deep learning systems can be exploited to optimize run times.

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

Interfacing living Neural cultures to control a Robot

This project is part of the NTNU Cyborg project and related research, -see:

Faglærer: Gunnar Tufte     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

Investigating New GPU Features for Performance (NVIDIA Volta)

Look into how effective are current optimization techniques for GPUs on the newest platforms, such as Jetson TX2, GTX 1080Ti and TeslaP100, and also explore new techniques for the recently announced Volta GPU from NVIDIA that inlcudes tensor processors.

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

Knowledge exploration in public linked data ontologies

Online ontologies like WikiData and Dbpedia are vast knowledge bases that provide structured information about millions of entities. They are specified in RDF(S) according to the principles of linked open data and are maintained by large online communities.

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

Kroppsbaserte bevegelsessensorer for 3D spill

I dette prosjektet skal det utvikles og evalueres løsninger for å koble kroppsbaserte bevegelsessensorer til 3D-spill plattformen Unity via Bluetooth LE. Målet er å kunne lese i sanntid fra f.eks. 8 forskjellige bevelgelsessensorer på kroppen (hender, armer, ben, torso), og bruke dette som input i 3D spill. En god kandidat for hardware er TIs nye CC2650 BLE SensorTag. Unity HW er Android eller Windows.

Faglærer: Dag Svanæs     Status: Tildelt     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

Language Technology Shared Task Competition

The master thesis work would involve participation in one of the "shared task competitions" in the language processing field, where training and test data is made available by the organisers. Examples include the Plagiarism Detection and Native Language Identification shared tasks mentioned in other master thesis proposals, but there are several others along the same line, for example, Grammatical Error Correction where the participating systems are given short English texts written by non-native speakers of English and should detect the grammatical errors present in the input texts (and return the corrected texts). Another example is Coreference Resolution where the task is to process a sentence like “She had a good suggestion and it was unanimously accepted" and mark the coreference (identical reference) between “a good suggestion” and “it”. Or a Statistical Machine Translation shared task where a common framework (including a baseline system) is provided and the goal just is to improve current translation methods in one of several possible ways.

An initial part of the thesis work would thus be to decide on which particular shared task to participate in (or whether data from a previous shared task should be used instead).

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

LANGUAGE TECHNOLOGY: ChatBots - Dialog interfaces - Text / Phone

It is now 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 7352 1290.

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

Machine learning for part-of-speech (POS) tagging

Part-of-speech tagging is the task of classifying words according to part-of-speech (e.g., noun, verb, etc.). It is a task very much suitable for the application of (and comparison of) a range of machine learning algorithms. The thesis topic could thus involve, e.g., the using tools such as the Memory-based tagger (MBT), Trigrams 'n Tags (TnT) or TreeTagger to build a POS tagger for a language such as Norwegian and/or comparing and combining several approaches (e.g., Support Vector Machines, Hidden Markov Models and Conditional Random Fields) for building part-of-speech taggers for under-resourced languages (such as a Scandinavian or an African language). The produced taggers should be compared to already available ones, such as the Oslo-Bergen Tagger for Norwegian.

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

Manuell oppfølging av sau på beite

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

Medical Image Computing and Deep Learning (DL)

Several projects related to this topic are available (in collaboration with St Olavs Hospital, SINTEF Medical Technology and ISB at the MH faculty at NTNU):

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

Merkenavnanalyse\puls

Hva sies det om SpareBank 1 SMN i media, facebook, twitter etc? Ved hjelp av sentimentanalyse kan vi se om vi omtales mye eller lite, og om det er positiv eller negativt. Er det mulig å lage en pulsmåler som gir oss temperaturen på det som sies om Sparebank 1? Vi har per i dag ulike løsninger som gir leter frem innhold hvor banken nevnes, men vi ønsker å finne en løsning som kan tallfeste innholdet og gi oss bedre statistikk til analyser og oppfølging

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

Mobile app: Integrated financial, energy and transport services

Transport, energy and housing represent 50% of the average household's expenses. They are also strongly interrelated. Where you live determines how much you need to commute and what transport means are available. It can mean having to buy an extra car or maybe cycling to work. The type of building also determines how much energy you will consume etc.
Together with partners from the Netherlands and Austria SINTEF is developing integrated financial, energy and transport services that will help users take better informed decisions based on personalized analyses and advice given through an app. The app will serve as an extension of banking services. Using household energy consumption data and users' moving patterns it will suggest to users how to improve their financial situation by saving money on energy and housing and choosing the right transportation means.
The app will be tested with users from three European cities, Helmond (Netherlands), Steinkjer (Norway) and Weiz (Austria).
If you feel like using your creativity and participate in shaping these services for Android, iPhone or both platforms, contact us asap.

This project will be done in collaboration with SINTEF Energy Research with 2 co-supervisors.

 

Faglærer: Sobah Abbas Petersen     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

Multi-Z Interpretation Autotracking

This project is associated with computational sciences activity group under the collaboration framework between Schlumberger, the largest oilfield services company in the world, and IPT (Petroleum Engineering & Applied Geophysics) and IDI (Computer and Information Science) at NTNU. The aim is to delineate (semi)-automatically the boundary of the geometric complex geobody (e.g. salt bodies) within a seismic cube. In the image/geometry processing world this task is equivalent to the extraction of 3D point clouds from a set of non-random images.

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

Named Entity Recognition in Norwegian Tweets

Named Entity Recognition (NER) plays a crucial role in many Natural Language Processing and Information Retrieval tasks, such as document search, clustering, information extraction, etc. This task is specially challenging when performed on tweets due to their noisy nature, such as non-standard spelling and grammar, code-switching and informal or unstructured text. When considering the multilingual nature of Twitter, the lack of resources for Scandinavian languages also produces additional challenges. In this project the students will explore current challenges to perform NER on Norwegian tweets. The main tasks in this project are the compilation of a corpus of Norwegian tweets with information about their entities and then to apply different machine learning algorithms in order to train the recognizer. One of the approaches to be explored consists in benefiting from entity linking with knowledge bases like Wikipedia in order to find the entities in these tweets.

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

Novelty detection

"Novelty detection" går ut på å avgjere om eit nytt dokument inneheld
informasjon som ikkje allereie er representert i dokument ein allereie
har i tekst-samlinga. Oppgåva går ut på å studere metoder for "novelty
detection", implementere nokre av teknikkane, og evaluere desse.

Faglærer: Kjetil Nørvåg     Status: Tildelt     Egnet for: Gruppe     Lenke: plink

OFF-Screen Rendering Toolbox

This project aims at the creation of an off-screen rendering toolbox that enables the user to create a collection of predefined renderings for one (or many) dataset(s).
The ability to extract renderings of a 3D model is crucial in many 3D object retrieval and recognition tasks. Furthermore, the increased use of Convolutional Neural Networks in such tasks makes the aforementioned problem even more important. The renderings can vary from simple planar to panoramic views to more complex examples of angular distance rendering and/or filtered outputs.

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

Online news ecosystems

The news media industry is experiencing a transformation from tightly controlled paper products to web-based services that include in-house news production, user-generated content and material from various news sources. Both the news content and the ads may be personalized to make them more relevant to individual readers. This transformation means that traditional media houses need to adopt new technology and work closer together with software providers and their own readers.

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

Optimizing Bio-Inspired Propulsion Systems using Genetic Algorithms

Biomimetic propulsion systems show great promise as more efficient substitutions for regular boat propellers. But to get high quality results, the movement pattern of said propulsion system must be optimized. At the moment, the most used way of determining movement patterns is by using mathematical models and attempting to reverse engineer fish and sea mammals.

Faglærer: Kazi Shah Nawaz Ripon     Status: Tildelt     Egnet for: En student     Lenke: plink

Parallel Go App -- Learning Parallel Programming in a Virtual World

Create a virtual world that may or may not interact with the real world, where you hunt down factoids on parallel programming much in the spirit of how you hunt down Pokemons in the recent Pokemon Go game.

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

Personalization - Natural Language Information Query (dialog)

What is the most natural way to get information about bus scedules or other well organised and structured data?

Faglærer: Rune Sætre     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

Prediction of problems in exposed fish farming operations

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.
 
Historical data from the EXPOSED Aquaculture SFI, in which NTNU IDI is a partner, (GPS position of vessels, time at net cage, wind information, distance to protective geography, time in different zones, interrupted operations, etc.), and NCE Seafood Innovation Cluster (AquaCloud) are available for research and innovation. Within EXPOSED there is ongoing work utilizing some of these data.
 
The topic for the project work and master thesis is to predict potential problems tied to transport, put out, feeding, sorting, delousing, treatment and slaughtering operations in exposed areas of fish farming based on available historical data (from various sensors, measurements of food consumption and growth, etc.). In addition, the analysis of videos showing fish behaviour should be considered. Collaboration with companies developing drones for collecting data in fish farms could be considered.
 
The thesis will be done at the Telenor-NTNU AI-Lab in connection with the EXPOSED Centre (dealing with monitoring and decision support systems) and NCE Seafood Innovation Cluster.
 
Supervisor: Kerstin Bach, NTNU will be main supervisor, with me (Agnar Aamodt) as an assitant on the side, given that I will be away on sabbatical in the coming academic year.
Co-supervisors: Sigmund Akselsen, Telenor, og Gunnar Senneseth, Sintef Ocean 

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

Prediction of problems in exposed fish farming operations

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.

Historical data from the EXPOSED Aquaculture SFI, in which NTNU IDI is a partner, (GPS position of vessels, time at net cage, wind information, distance to protective geography, time in different zones, interrupted operations, etc.), and NCE Seafood Innovation Cluster (AquaCloud) are available for research and innovation. Within EXPOSED there is ongoing work utilizing some of these data.

The topic for the project work and master thesis is to predict potential problems tied to transport, put out, feeding, sorting, delousing, treatment and slaughtering operations in exposed areas of fish farming based on available historical data (from various sensors, measurements of food consumption and growth, etc.). In addition, the analysis of videos showing fish behaviour should be considered. Collaboration with companies developing drones for collecting data in fish farms could be considered.

The thesis will be done at the Telenor-NTNU AI-Lab in connection with the EXPOSED Centre (dealing with monitoring and decision support systems) and NCE Seafood Innovation Cluster.

Supervisor: Kerstin Bach and Agnar Aamodt, NTNU
Co-supervisors: Sigmund Akselsen, Telenor, og Gunnar Senneseth, Sintef Ocean

 

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

Prediction of salmon lice outbreaks in fish farms

Outbreaks of salmon lice in fish farms are a main cause for stress, reduced growth and death in commercial fish farming. Such attacks have been estimated to account for losses in billions of NOKs each year in Norwegian farms. Even a minor reduction in outbreaks would result in high profitability gains.
 
NCE Seafood Innovation Cluster is currently developing the AquaCloud concept were data from various sources is collected and will be made available for research and innovation. One of the initiatives is focusing on the prediction of salmon lice outbreaks based on historical data. This is a data source that will be considered int his project. A pilot involving IBM Watson is part of their initiative, and will be one option to look into.
 
The topic for the project work and master thesis is to investigate how outbreaks of salmon lice attacks can be predicted based on available historical data (from various sensors, manual count of number of lice, measurements of food consumption and growth, etc.). In addition, the analysis of videos showing fish behaviour should be considered. Collaboration with companies developing drones for collecting data in fish farms could be considered.
 
The thesis will be done at the Telenor-NTNU AI-Lab with connections to the EXPOSED Aquaculture SFI, in which NTNU IDI is a partner, (dealing with monitoring and decision support systems), and NCE Seafood Innovation Cluster.

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

Prediction of salmon lice outbreaks in fish farms

Outbreaks of salmon lice in fish farms are a main cause for stress, reduced growth and death in commercial fish farming. Such attacks have been estimated to account for losses in billions of NOKs each year in Norwegian farms. Even a minor reduction in outbreaks would result in high profitability gains.

NCE Seafood Innovation Cluster is currently developing the AquaCloud concept were data from various sources is collected and will be made available for research and innovation. One of the initiatives is focusing on the prediction of salmon lice outbreaks based on historical data. This is a data source that will be considered int his project. A pilot involving IBM Watson is part of their initiative, and will be one option to look into.

The topic for the project work and master thesis is to investigate how outbreaks of salmon lice attacks can be predicted based on available historical data (from various sensors, manual count of number of lice, measurements of food consumption and growth, etc.). In addition, the analysis of videos showing fish behaviour should be considered. Collaboration with companies developing drones for collecting data in fish farms could be considered.

The thesis will be done at the Telenor-NTNU AI-Lab with connections to the EXPOSED Aquaculture SFI, in which NTNU IDI is a partner, (dealing with monitoring and decision support systems), and NCE Seafood Innovation Cluster.

Supervisor: Kerstin Bach and Agnar Aamodt, IDI, NTNU
Co-supervisors: Sigmund Akselsen, Telenor og Gunnar Senneseth, Sintef Ocean

 

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

Predikering/optimering i bysykkelanlegg

Bysykkel er midt i en eksplosiv vekst internasjonalt. De viktigste driverne er ny teknologi, økt urbanisering og behov for fleksibel og integrert mobilitet.

Bysykler gir brukerne mulighet til sykle mellom A og B, gitt at det er tilgjengelige sykler i A og ledige låser i B. For å redusere antall tomme og fulle stativer brukes gjerne servicebiler til å flytte sykler mellom de ulike stativene. Planleggingen av både nye stativlokasjoner og servicebilenes ruter gjøres manuelt og er ofte preget av tilfeldigheter. Våren 2018 åpnes det et nytt bysykkelanlegg i Trondheim.

Samarbeidspartneren til denne prosjekt-/masteroppgaven er Urban Infrastructure Partner (UIP). UIP er en norsk start-up som utvikler teknologi for bedre ressursutnyttelse i byer. Deres første prosjekt har vært en ny bysykkelordning i Oslo, som ble lansert våren 2016. De ønsker å utvikle og tilby smarte, bærekraftige og effektive løsninger, som hjelper byer med å takle deres mobilitetsutfordringer. I denne oppgaven vil det kunne jobbes med reelle data (historisk og sanntid) fra bysykkelsystemet i Oslo, og eventuelle modeller som utvikles vil kunne testes her.

Overordnet tema for oppgave blir å benytte maskinlæring, kunstig intelligens og optimeringsalgoritmer til å forbedre operasjonene i bysykkelanlegg. Flere ulike problemstillinger er aktuelle og den som blir tildelt denne oppgaven vil være med å utforme disse i samarbeid med UIP og veiledere. Eksempler:

Faglærer: Trond Aalberg     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

Programmerbare materialer

I dette prosjektet skal det utvikles verktøy som gjør det mulig for ikke-programmerere å programmere ansamlinger av små programmerbare enheter. Oppgaven er utforskende av natur, med fokus både på design, implementasjon og bruk. Det skal konkret arbeides med en ansamling av ca. 100 små 1x2 cm brikker med innebygget prosessor, LEDs og IR kommunikasjon. Det vil bli programmert i C. Dersom ønskelig kan det også bygges konkrete verktøy v.h.a. 3D-printer.

Faglærer: Dag Svanæs     Status: Tildelt     Egnet for: Gruppe     Lenke: plink

Reasoning with building information models

Building information models contain information about topology, geometry and semantics of buildings and they can be thought of as ontologies.

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

Research method in AI: Reproducibility of results

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 the 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: Tildelt     Egnet for: Gruppe     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

Security Testing of Web Frameworks

Today, many web applications are built on frameworks. Since frameworks provide useful abstraction for common functionalities they are thus attractive from cost and development time perspectives. However, from security perspectives an application built on a framework is only as secured as its underlining framework.

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

Skybasert Internet-of-Things (IoT) - evaluering av plattformer

En rekke av de store IT-firmaene leverer nå skybaserte IoT rammeverk og tjenester. Dette er plattformer som vil bli viktig i nye applikasjoner for alt fra helse til smarte byer og spill. I denne oppgaven skal de 3-4 mest aktuelle rammeverkene sammenlignes (Amazon, Google, Oracle, OpenSource Eclipse). Etter sammenligningen skal et referansecase implementeres i ett eller flere av rammeverkene. Caset er knyttet til helse, med flere trådløse sensorer og aktuatorer som kobler seg til nettet og videre til skytjenestene både direkte (IPv6/6LoWPAN) og via smarttelefon (Bluetooth LE/4G).

Faglærer: Dag Svanæs     Status: Tildelt     Egnet for: Gruppe     Lenke: plink

Smart health app

The purpose of this project is to develop IoT based application for collecting and analyzing movement data of post-stroke patients. The data collected and analyzed will be used by doctors to guide the patients for rehabilitation. 

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

Smart Image Segmentation

Background:

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

Social Innovation platforms - Co-supervisor Ilias O. Pappas

The main objective of SOCRATIC www.socratic.eu is to facilitate a platform for the citizens and/or organisations to collaboratively identify specific innovative solutions for achieving the desired Global Sustainability Goals, as defined by United Nations. The platform will allow individuals, collectives, institutions, companies or administration:

Faglærer: Maria Letizia Jaccheri     Status: Valgbart     Egnet for: Gruppe     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

Spam or spyware detection (text classification)

To apply one (or several to compare) automatic classification techniques (such as Self-Organizing Maps, k-Nearest Neighbour, Naïve Bayes, Expectation-Maximization, etc.) to different types of texts, e.g., to determine whether an incoming e-mail is spam, whether a user license agreement includes spyware, or which type of news a specific news item belongs to.

 

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

Style identification of authors and newspapers (COPY)

Different newspapers has different styles of writing about events.  
In particular columnists tend to have different opinions or perspectives about different events. The aim of this project is to identify the styles of different authors by analyzing the articles written by them, following the style identification of different newspapers.

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

Surgery Scheduling using Heuristic Approach

Surgery scheduling is a complex task and it is usually performed manually in two separate steps. In the first step, advance scheduling, surgeries are assigned a date. In the second step, allocation scheduling, surgeries are allocated to a specific operation room (OR) and time slot. This two-step manual approach is time-consuming and often does not provide optimal result. Heuristic/Bio-Inspired approaches can solve this problem efficiently. The goal of this research is to discover an overall efficient solver for the general surgery scheduling problem by combining recent Heuristic/Bio-Inspired methods to overcome these challenges.

Faglærer: Kazi Shah Nawaz Ripon     Status: Valgbart     Egnet for: En student     Lenke: plink

The Little Doormaid - an interactive a fairy tale about social innovation

download The Little Doormaid is a fairy tale about social innovation. The practical goal of this project is to develop and test a prototype of an interactive experience that conveys the message of the fairy tale. Development encompasses the choices of which technology and which hardware and software to use. The scientific goal of this work is to produce 1) a state of the art description of related efforts; 2) relevant research questions; 3) evaluation of the produced artefact. More information here  

Faglærer: Maria Letizia Jaccheri     Status: Valgbart     Egnet for: En student     Lenke: plink

The use of service robots in elderly care

The number of elderly who need care services is increasing in most countries. This results in new ways of organizing and delivering care services. Technology is playing an increasing role in this respect. Various types of technology such as smart house, internet of things, and tracking devices are used to provide care and enable independent living for elderly.

Faglærer: Babak A. Farshchian     Status: Valgbart     Egnet for: Gruppe     Lenke: plink

The Virtual Hospital of the Future: an arena for virtual collaboration and training

Several projects related to this topic are available:

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

Topic modelling vs keyword extraction in terms of forming document search queries

RQ: Which of the two, or a combination, is better when it comes to summarizing a text into keywords to be used in search queries?

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

Travel patterns to and from picked relevant locations in a region day by day

The main goal with this project is to develop a tool to make it possible to analyse and estimate the number of people traveling to and from selected relevant locations in a region day by day. More specifically, the project is to develop a way to provide estimate of catchment areas of people traveling within the region; identify massive events and their effects on travel patterns; find correlations between opening/closing of transport infrastructure. The project is done in collaboration with Telenor that will provide anonymous location (xDR) data (frequency of measurements every 5 minutes or shorter intervals if required) combined with data corresponding to roads/railways/etc.

 

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

Trådløs plattform for digitale fysiske lagspill

Bluetooth LE og ANT+ System on Chip, f.eks. fra Nordic Semiconductor i Trondheim, muliggjør trådløs kommunikasjon mellom mange billige små programmerbare enheter. Dette kalles ofte Internet of Things (IoT). IoT åpner opp for en rekke spennende anvendelser bl.a. innen sport og spill. I dette prosjektet skal det utvikles en plattform for digitalt forsterkede lagspill. Ideen er å kunne utstyre brukere med digitale armbånd, leggbånd, samt lage baller og faste punkter som kan programmeres av vanlige brukere via en mobiletelefon app til å endre farge og gi lyd ved berøring eller bevegelse. Det kreves god kunnskap i programmering og evne til å sette seg inn i nye teknologier. Oppgaven egner seg for tre studenter, en som jobber med berøringssensorer, en som jobber med trådløs kommunikasjon og en som jobber med visuell programmering av oppførsel.

Faglærer: Dag Svanæs     Status: Valgbart     Egnet for: Gruppe     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

Utilizing Big Data for Inflow and Outflow Analysis for ensuring impact in Tourism Industry

[Project in collaboration with Telenor-NTNU AILab]

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

Web Interface for Deep Learning with Case Study in Facial Recognition

Below is our proposal for the Master's thesis:
​​Website idea:
Run Python scripts for deep learning.
Background: Python is a popular library for deep learning because of all the libraries that use it, e.g. Theano, TensorFlow, Keras, Lasagne. A deep learning script typically loads data, pre-processes it, sets up a model, specifies an objective/loss function, and applies it in a training loop to train the network.
​Typically, on the website, you would be able to upload your script and run it.
For archival reasons it would be nice if the website could take a backup of the Python script after it as been uploaded.
Using callback mechanisms be able to listen to the training loop which then writes training loss, accuracy etc. The website would then be able to read these text files and plot graphs.
This, along with a backup of the Python script, make it easier to monitor the training and look up past experiments.
Come up with a file structure for handling backups and text files used for plotting, e.g. a tag-based approach may make it easy to search for similar past experiments.
A separate tool that can create binary blobs of data for when the data is too big for a GPU. Common solutions are HDF5 and LMDB.
A tool for visualisation data from a binary blob should also be present.
Visualisation of outputs for each layer in an artificial neural network model.
Other visualisation techniques, such as salient analysis, deep dreaming, and de-convolution.
Face recognition with expression data
Three questions that can be solved by the thesis:
Does face expression help with face recognition?
What method is best for incorporating face expression in face recognition software?
Baseline: ​Face recognition without face expression, i.e. face expression does not add anything of value.
Face recognition model has an extra input representing the face expression of the input data. The face expression data can be learned from a different dataset.
Face recognition model takes face data and produces two outputs: (i) subject, and (ii) face expression. In other words, the model needs to learn a combined representation.
This last point would require a dataset with both subject and face expression labels for training. This means that you would have to create a dataset.
​Does the number of available face expressions affect the face recognition performance?
This means repeating experiments on a different number of facial expressions. Keep in mind that there are many ways to measure the performance of a model.

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

Web security testing using generative testing method

In January 2014, news outlets reported that the usernames and passwords of 4.6
million Snapchat users had leaked, when intruders exploited critical security
holes in Snapchat's "non-public" back-end API. While such catastrophic
incidents are uncommon, security bugs that could allow critical data leaks are
regularly discovered in all kinds of web APIs. Most such bugs are simple to
fix (read: patch) once discovered, and many could have been discovered with
somewhat more rigorous security testing. In this project we will investigate
how to use *generative* testing techniques to find such bugs.

In generative testing, a tester does not explicitly write individual test
cases. Rather, the tester provides properties specifying how a software
component is supposed to behave, and then a software library generates and
executes random test cases asserting these properties. The best known
generative testing library is QuickCheck [1], which targets Haskell.

It is not clear how best to specialize the existing general-purpose generative
testing techniques to find security bugs in web APIs (that is what we are
going to figure out!), but some ideas for things to include into a specialized
system are
- knowledge about typical security flaws seen in web APIs,
- knowledge about authorization and authentication protocols, and
- comprehension of formal languages used to describe custom data types used
on the web. (Most likely, we need to understand and be able to fuzz data
from JSON Schemas or XML Schemas.)
The bulk of the student's work will be to ponder about, experiment with, and
evaluate many such ideas for specializing generative testing techniques for
security testing of web APIs.

The project will have three supervisors, two external to NTNU. The two external supervisors are Bjarte M. Østvold, who is head of the Information security group at
the Norwegian Computation Center (NR) and Edvard K. Karlsen, who is a consultant
at Kantega.

[1]: Claessen and Hughes, "QuickCheck: a lightweight tool for random
testing of Haskell programs", ICFP 2000.

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

[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

“Information as commodity” in the Internet of Things (IoT)

Co-supervisor: Hai Thanh Nguyen (Telenor)

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