Faculty of ICT
B.Sc. I.T. (Hons.) in Artificial Intelligence
ICT3909 (Final Year Project in Artificial Intelligence
– 30ECTS)
Proposal Form
Title:
AR Temples
Project Supervisor:
Alexiei Dingli
Project Co-supervisor:
(if applicable)
Main Subject Area/s:
Vanessa Camilleri
Intelligent Interfaces, Creative Technologies, etc.
The idea behind this project is to create an Augmented Reality framework
(similar to http://www.motive.io/) capable of creating animated 3D
environments. The framework should allow people to create their own AR
environments inclusive of sounds using an AR editor and Geo-Locations. The
AR used is marker less so it either recognizes existent structures of GPS
positions. The user can also interact with the AR in a similar way as Motive.io
does.
Brief Project Description inc.
References:
Once created, users can download the AR information about the site on their
(word limit approx. 300 words)
mobile device and view the AR experience. An important thing to note is that
some sites might contain different layers of AR information. E.g. the same site
might include a Neolithic Temple, a Phoenician Temple and a Roman Temple.
In this case the user has to also provide timelines so that users can navigate
through space and time easily.
The framework should work as a web interface but the AR interface should be
deployed through an APP on a mobile device.
Resources Required:
Android SDK
Recommended Prerequisites /
Knowledge Required and
rd
Supporting 3 Year Study -
Intelligent Interfaces Course
units:
Foreseeable Ethical Issues and
How these will be tackled:
No
(if applicable)
Deadline: Monday 25th April 2016. To be submitted to Ms Francelle Scicluna
(francelle.scicluna@um.edu.mt)
Faculty of ICT
B.Sc. I.T. (Hons.) in Artificial Intelligence
ICT3909 (Final Year Project in Artificial Intelligence
– 30ECTS)
Proposal Form
Title:
Gamified Beacons
Project Supervisor:
Alexiei Dingli
Project Co-supervisor:
(if applicable)
Main Subject Area/s:
Intelligent Interfaces, Creative Technologies, etc.
Beacon technology is becoming extremely important in today’s world because
it can be used to localize goods, provide indoor navigation and much more.
The scope behind this FYP is to create a supermarket experience using a
mobile app and beacons. The app is expected to have the following minimum
features (but it can have others):
Brief Project Description inc.
References:
(word limit approx. 300 words)
1. Provide a shopping list for the customers.
2. Provide navigation through the supermarket based upon the shopping
list (and biased towards promotions of the supermarket).
3. Provide recommendations (based upon the ones in the shopping list)
4. Allow the supermarket to send targeted and localized promotions to the
user.
5. Create gamification elements (E.g. games while people are queuing to
pay whilst getting discounts if they win the game on the final bill)
6. Electronic loyalty card system
Resources Required:
Beacons, Mobile Device, Android SDK
Recommended Prerequisites /
Knowledge Required and
Supporting 3rd Year Study units:
Intelligent Interfaces Course
Faculty of ICT
B.Sc. I.T. (Hons.) in Artificial Intelligence
ICT3909 (Final Year Project in Artificial Intelligence
– 30ECTS)
Proposal Form
Title:
OpenTTD Deep AI
Project Supervisor:
Alexiei Dingli
Project Co-supervisor:
(if applicable)
Main Subject Area/s:
Machine Learning
The project involves designing and developing a new AI for Open Transport
Tycoon Deluxe (https://www.openttd.org). The student can build on another
project performed this year.
However, the difference being that the student is expected to use Deep AI.
Deep learning has been successfully developed during the last decade and its
constantly gaining importance. A few months ago, Google DeepMind algorithm
Brief Project Description inc.
called Alpha Go (https://www.tastehit.com/blog/google-deepmind-alphago-how-
References:
it-works/) managed to beat the world champion in Go. To do so, it made heavy
(word limit approx. 300 words)
use of deep learning together with other technologies. Go is considered as one
of the most complex games available because the number of possible games
in Go is bigger than all the atoms in the universe.
Because of this, the student is expected to research into the workings of
AlphaGo and create a Deep AI system capable of playing OpenTTD using
deep learning techniques.
Resources Required:
Online connection
Recommended Prerequisites /
Knowledge Required and
rd
Supporting 3 Year Study -
Machine Learning
units:
Foreseeable Ethical Issues and
How these will be tackled:
No
(if applicable)
Deadline: Monday 25th April 2016. To be submitted to Ms Francelle Scicluna
(francelle.scicluna@um.edu.mt)
Faculty of ICT
B.Sc. I.T. (Hons.) in Artificial Intelligence
ICT3909 (Final Year Project in Artificial Intelligence
– 30ECTS)
Proposal Form
Title:
Virtual and Augmented Reality Framework for Kids
Project Supervisor:
Alexiei Dingli
Project Co-supervisor:
(if applicable)
Main Subject Area/s:
Intelligent Interfaces
The scope of this project is to create an augmented-reality and virtual-reality
authoring environment for children. It should be an extension
(https://github.com/LLK/scratchx/wiki) of MIT’s Scratch project
(https://scratch.mit.edu/). This environment should allow children to create
experiences that mix real and virtual elements or an environment, which create
a total immersion for the child. Children can display virtual objects on a realworld scene observed through a video camera, and they can control the virtual
Brief Project Description inc.
References:
(word limit approx. 300 words)
world through interactions between physical objects. This project aims to
expand the range of creative experiences for young authors, by presenting AR
and VR technologies in ways appropriate for this audience. In this process, the
student should investigate how young children conceptualize augmented and
virtual reality experiences, and shape the authoring environment according to
this knowledge.
The final deliverable should be a scratch extension, which also allows children
to deploy their finished product onto a mobile device with ease.
As a start have a look at (http://ael.gatech.edu/lab/research/authoring/arspot/).
Resources Required:
Mobile Device
Recommended Prerequisites /
Knowledge Required and
rd
Supporting 3 Year Study -
Intelligent Interfaces
units:
Foreseeable Ethical Issues and
How these will be tackled:
No
(if applicable)
Deadline: Monday 25th April 2016. To be submitted to Ms Francelle Scicluna
(francelle.scicluna@um.edu.mt)
Faculty of ICT
B.Sc. I.T. (Hons.) in Artificial Intelligence
ICT3909 (Final Year Project in Artificial Intelligence
– 30ECTS)
Proposal Form
Title:
Interactive Painting
Project Supervisor:
Alexiei Dingli
Project Co-supervisor:
(if applicable)
Main Subject Area/s:
Claudia Borg
Intelligent Interfaces, Conversational Agents
The idea behind the Interactive Painting is to make use of an existent famous
portrait (such as the Mona Lisa) and make her talk. The student needs to
create a sophisticated chat bot capable of engaging in a meaningful
conversation with the user. Students are encouraged to use existent
frameworks such as ChatScript, the open-source Natural Language scripting
language and engine running bots, which successfully managed to win the
Loebner Prize. The main challenge is to make these engines run on a mobile
device.
Brief Project Description inc.
References:
The system should also make use of the camera on the mobile device in order
(word limit approx. 300 words)
to identify people and make the painting interact with the people and the
environment in front of it.
The final deliverable in this case is a painting running from a mobile device and
capable of conversing with the users. The topics discussed should be related to
the environment in which it is placed and the painting i.e. the context in which it
was painted, the artist, the subject of the painting, etc. Such a project will be
used in schools in order to teach children about art since they can easily start a
conversation with the agent.
Resources Required:
Mobile Device
Recommended Prerequisites /
Knowledge Required and
rd
Supporting 3 Year Study -
Intelligent Interfaces, NLP
units:
Foreseeable Ethical Issues and
How these will be tackled:
No
(if applicable)
Deadline: Monday 25th April 2016. To be submitted to Ms Francelle Scicluna
(francelle.scicluna@um.edu.mt)
Faculty of ICT
B.Sc. I.T. (Hons.) in Artificial Intelligence
ICT3909 (Final Year Project in Artificial Intelligence
– 30ECTS)
Proposal Form
Title:
Augmented Reality Autistic Experience
Project Supervisor:
Alexiei Dingli
Project Co-supervisor:
(if applicable)
Main Subject Area/s:
Vanessa Camilleri
Intelligent Interfaces
The idea behind this project is to create an Augmented Reality experience
which mimic the feelings of an Autistic Person as can be seen in the following
video : https://www.youtube.com/watch?v=KmDGvquzn2k
This project will be held in conjunction with other departments of the University
Brief Project Description inc.
References:
(word limit approx. 300 words)
of Malta which specialize in Autism. The idea behind this project is to place
those students studying to become teachers in the shoes of an Autistic child.
By doing so, the prospective teachers do not only read about autism but they
can also experience it thus being in a better position to understand their
children.
We will also include some gamification elements within the experience thus
making the prospective teachers go through normal tasks as an autistic child
and the difficulties, which these children pass through on a daily basis.
Resources Required:
Mobile Device
Recommended Prerequisites /
Knowledge Required and
rd
Supporting 3 Year Study units:
Intelligent Interfaces
Foreseeable Ethical Issues and
How these will be tackled:
The research will be held with consensual adults.
(if applicable)
Deadline: Monday 25th April 2016. To be submitted to Ms Francelle Scicluna
(francelle.scicluna@um.edu.mt)
Faculty of ICT
B.Sc. I.T. (Hons.) in Artificial Intelligence
ICT3909 (Final Year Project in Artificial Intelligence
– 30ECTS)
Proposal Form
Title:
PenTaGon: Personal Task Data Generator
Project Supervisor:
Charlie Abela
Project Co-supervisor:
(if applicable)
Main Subject Area/s:
Chris Staff
Knowledge discovery, Information Retrieval, Web Intelligence
The research field of personal information management (PIM) deals with how
people acquire, organise, maintain, retrieve, use and control the distribution of
information items such as documents, web pages and emails for everyday use
to complete tasks. One of the main goals behind PIM is the design and
evaluation of tools intended to support users in dealing with fragmented
information.
Evaluating PIM tools requires massive amounts of data to be collected,
Brief Project Description inc.
however this is problematic since people have great concerns when it comes to
References:
relinquishing their browsing data, for example, because this is considered to be
(word limit approx. 300 words)
private
and
confidential.
Furthermore,
when
such
data
is
collected
anonymously, it might still not be possible to publically share it.
In order to overcome this problem, this research will first focus on the concept
of tasks and will consider different task classification approaches. Secondly, it
involves the development of PenTaGon, a synthetic task data generator that
allows for the flexible creation of synthetic task data with realistic
characteristics. PenTaGon will also provide need to allow task data to be
visualised, compared and analysed.
Resources Required:
Personal Computer
This research builds upon material covered in:
Recommended Prerequisites /
Knowledge Required and
Supporting 3rd Year Study -
•
ICS2205 – Web Intelligence
rd
Recommended 3 Year Study Units:
units:
•
ICS3208 – Advanced Web Intelligence
•
ICS3207 – Knowledge Discovery and Management
Foreseeable Ethical Issues and
How these will be tackled:
N/A
(if applicable)
Deadline: Tuesday 26th April 2016. To be submitted to Ms Francelle Scicluna
(francelle.scicluna@um.edu.mt)
Faculty of ICT
B.Sc. I.T. (Hons.) in Artificial Intelligence
ICT3909 (Final Year Project in Artificial Intelligence
– 30ECTS)
Proposal Form
Title:
dOMiNiuM (or dOMiNiuM PubliCuM): Opinion mining of new portal comments
Project Supervisor:
Dr. Charlie Abela
Project Co-supervisor:
(if applicable)
Main Subject Area/s:
Dr. Joel Azzopardi
Knowledge Discovery, Text Processing, Web Intelligence
News portals nowadays allow their readers to voice their opinions through the
comments sections associated with each issued article. In this regards different
blog comment hosting services for web sites and online communities such as
Disqus provide APIs through which comments can be extracted.
The field of opinion mining refers to the application of natural language
processing, computational linguistics, and text analytics to identify and extract
Brief Project Description inc.
References:
(word limit approx. 300 words)
subjective information in source materials. Research has focused on identifying
sentiment in tweets and other online resources such as news articles,
comments and blogs.
The main objective behind this project is to develop a web-based tool called
dOMiNiuM PubliCuM, that allows for the analyzes of news portal comments
and identify the general opinion about specific articles. The tool should
potentially also focus on tracking the opinions of the same user/s over time in
relation to similar news items. dOMiNiuM PubliCuM should allow for real-time
reporting, analytics and visualizations to track the evolving trends (opinions)
about an article.
Resources Required:
Personal Computer
This research builds upon material covered in:
Recommended Prerequisites /
Knowledge Required and
Supporting 3rd Year Study -
•
ICS2205 – Web Intelligence
rd
Recommended 3 Year Study Units:
units:
•
ICS3208 – Advanced Web Intelligence
•
ICS3207 – Knowledge Discovery and Management
Foreseeable Ethical Issues and
How these will be tackled:
N/A
(if applicable)
Deadline: Monday 25th April 2016. To be submitted to Ms Francelle Scicluna
(francelle.scicluna@um.edu.mt)
Faculty of ICT
B.Sc. I.T. (Hons.) in Artificial Intelligence
ICT3909 (Final Year Project in Artificial Intelligence
– 30ECTS)
Proposal Form
Title:
Aggression detection in urban areas based on audio analysis
Project Supervisor:
Dr. George Azzopardi
Project Co-supervisor:
(if applicable)
Main Subject Area/s:
Computer vision, pattern recognition, data mining, machine learning
Aggression detection in urban environments is an important application as it
increases the piece of mind of residents. Surveillance cameras are used to
record the visual and audio information of a particular place. This topic has
attracted the interest of a large research community.
The aim of this project is to detect and recognize audio signals, which are
Brief Project Description inc.
associated with aggression. In particular we will investigate three types of audio
References:
signals, namely gunshots, screams, and breaking of glass. We will use a
(word limit approx. 300 words)
benchmark data set of audio signals created by the University of Salerno with
whom I have close collaboration.
For the detection of the mentioned three patterns we will investigate signal
processing and computer vision techniques, namely dynamic time warping and
COSFIRE filters.
This work has a high potential of a publication.
Resources Required:
Recommended Prerequisites /
Knowledge Required and
Supporting 3rd Year Study units:
N/A
No prerequisites
Supporting 3rd year study units: ICS3206, ICS3129
Foreseeable Ethical Issues and
How these will be tackled:
N/A
(if applicable)
Deadline: Monday 25th April 2016. To be submitted to Ms Francelle Scicluna
(francelle.scicluna@um.edu.mt)
Faculty of ICT
B.Sc. I.T. (Hons.) in Artificial Intelligence
ICT3909 (Final Year Project in Artificial Intelligence
– 30ECTS)
Proposal Form
Title:
Automatic texture recognition
Project Supervisor:
Dr. George Azzopardi
Project Co-supervisor:
(if applicable)
Main Subject Area/s:
Computer vision, image processing, machine learning
Texture is an important feature that can help to segment objects from their
background. For instance, Figure 1 shows an example of cheetah surrounded
by golden grass. If we only rely on the contours of the objects we will not be
able to segment (i.e. to delineate the object) from the background. In such
examples objects can be segmented by also considering the texture.
Brief Project Description inc.
References:
(word limit approx. 300 words)
Figure 1 - Examples of fingerprints. Each row shows the prints of the finger in different
arrangements.
The aim of this project is to investigate a novel algorithm called CORF with
Push-Pull inhibition for texture analysis. The said algorithm has already been
found to achieve state-of-the-art performance in contour detection but its
flexibility allows it to be suitable also for texture analysis. For evaluation
purposes we will use a benchmark data set of texture, namely Kylberg Texture
Database: http://tinyurl.com/jn4twuk
There is a high potential that this project will lead to a publication.
Downloads:
• Matlab code of CORF with Push-Pull inhibition: http://tinyurl.com/kkkplnd
• Research paper: http://tinyurl.com/pexz4xh
Resources Required:
Recommended Prerequisites /
Knowledge Required and
Supporting 3rd Year Study units:
N/A
No prerequisites
Supporting 3rd year study units: ICS3206, ICS3129
Foreseeable Ethical Issues and
How these will be tackled:
N/A
(if applicable)
Deadline: Monday 25th April 2016. To be submitted to Ms Francelle Scicluna
(francelle.scicluna@um.edu.mt)
Faculty of ICT
B.Sc. I.T. (Hons.) in Artificial Intelligence
ICT3909 (Final Year Project in Artificial Intelligence
– 30ECTS)
Proposal Form
Title:
Biometric analysis for personal identification using retinal fundus images
Project Supervisor:
Dr. George Azzopardi
Project Co-supervisor:
(if applicable)
Main Subject Area/s:
Biometrics, security, computer vision, image processing
It has been known since 1935 that the vessel structure of the retinas of human
beings is unique. Even the retinas of identical twins have different pattern. This
property makes it ideal for retinal images to be used in biometric systems for
the identification of persons.
The goal of this project is to develop a person identification system based on
the matching of retinal images. We will investigate feature matching techniques
such as SIFT, SURF, LBP among others. These techniques are not only
important for this application but they are used in various computer vision
applications. For the evaluation of the project we will use real images that we
will obtain from Mater Dei.
Brief Project Description inc.
The delineation of vessels in retinal images is a crucial step in medical
References:
differential analysis. The ability of delineating the vessel tree from the rest of
(word limit approx. 300 words)
the image can help medical experts to perform better diagnosis. Fig. 1b shows
an example of the expected output to a given coloured retinal fundus image.
There is a high potential that this project will lead to a publication.
Figure 1- Example of a retinal image taken from http://tinyurl.com/jlcdtvx.
Resources Required:
Recommended Prerequisites /
Knowledge Required and
Supporting 3rd Year Study units:
N/A
No prerequisites
Supporting 3rd year study units: ICS3206, ICS3129
Foreseeable Ethical Issues and
How these will be tackled:
N/A
(if applicable)
Deadline: Monday 25th April 2016. To be submitted to Ms Francelle Scicluna
(francelle.scicluna@um.edu.mt)
Faculty of ICT
B.Sc. I.T. (Hons.) in Artificial Intelligence
ICT3909 (Final Year Project in Artificial Intelligence
– 30ECTS)
Proposal Form
Title:
Delineation of vessels in retinal fundus images
Project Supervisor:
Dr. George Azzopardi
Project Co-supervisor:
(if applicable)
Main Subject Area/s:
Medical image analysis, computer vision, image processing
Retinal images provide a unique opportunity to investigate the health status of
a person in a non-invasive way. Besides eye-related diseases (e.g. diabetic
retinopathy, and age-macular diseases), from such images medical experts
can identify signs of other pathologies such as atherosclerosis and
hypertension.
The delineation of vessels in retinal images is a crucial step in medical
differential analysis. The ability of delineating the vessel tree from the rest of
the image can help medical experts to perform better diagnosis. Fig. 1b shows
an example of the expected output to a given coloured retinal fundus image.
Brief Project Description inc.
References:
(word limit approx. 300 words)
Figure 1- Example of a (a) coloured retinal fundus image and the corresponding vessel
segmentation that is hand drawn by an expert.
This is not a new problem in the literature and several attempts have been
made. Recently, a vessel delineation algorithm, called B-COSFIRE, has been
proposed for the delineation of vessels in retinal images (Azzopardi et al.
2015). While it is highly effective to detect isolated vessels, it suffers from
insufficient ability to segment regions where two or more vessels cross each
other. These regions are called bifurcations and crossovers.
The aim of this project is to combine bifurcation and crossover detectors with
the existing vessel segmentation B-COSFIRE algorithm in to improve the
robustness of the algorithm. Besides medical image analysis, such a
segmentation algorithm is important for many other computer vision
applications.
There is high potential that this project will lead to a publication.
Downloads:
• Matlab code of the B-COSFIRE algorithm: http://tinyurl.com/pfxcxl5
• Research paper of the B-COSFIRE algorithm: http://tinyurl.com/l43z7oy
Resources Required:
Recommended Prerequisites /
Knowledge Required and
N/A
No prerequisites
rd
Supporting 3 Year Study units:
rd
Supporting 3 year study units: ICS3206, ICS3129
Foreseeable Ethical Issues and
How these will be tackled:
N/A
(if applicable)
Deadline: Monday 25th April 2016. To be submitted to Ms Francelle Scicluna
(francelle.scicluna@um.edu.mt)
Faculty of ICT
B.Sc. I.T. (Hons.) in Artificial Intelligence
ICT3909 (Final Year Project in Artificial Intelligence
– 30ECTS)
Proposal Form
Title:
Fingerprint recognition
Project Supervisor:
Dr. George Azzopardi
Project Co-supervisor:
(if applicable)
Main Subject Area/s:
Biometrics, security, computer vision, image processing
Fingerprints are among the most popular features that are used in biometric
systems for the identification of persons. For instance, every person that goes
to USA has his finger prints recorded in their databases.
In this project we will use the FVC2000 benchmark data sets of fingerprints:
http://bias.csr.unibo.it/fvc2000/download.asp. Figure 1 shows examples of real
fingerprints taken from this data set.
We will investigate the B-COSFIRE algorithm that is very effective in the
segmentation of ridges together with other computer vision algorithms that can
be used for the matching of fingerprints. The B-COSFIRE algorithm has never
been used in this application. I expect that with its robustness to noise
(disconnected ridges), there is a good chance to improve the state-of-the-art.
Brief Project Description inc.
References:
(word limit approx. 300 words)
Figure 1 - Examples of fingerprints. Each row shows the prints of the same finger in
different arrangements.
There is a high potential that this project will lead to a publication.
Downloads:
• Matlab code of the B-COSFIRE algorithm: http://tinyurl.com/pfxcxl5
• Research paper of the B-COSFIRE algorithm: http://tinyurl.com/l43z7oy
Resources Required:
Recommended Prerequisites /
Knowledge Required and
Supporting 3rd Year Study units:
N/A
No prerequisites
Supporting 3rd year study units: ICS3206, ICS3129
Foreseeable Ethical Issues and
How these will be tackled:
N/A
(if applicable)
Deadline: Monday 25th April 2016. To be submitted to Ms Francelle Scicluna
(francelle.scicluna@um.edu.mt)
Faculty of ICT
B.Sc. I.T. (Hons.) in Artificial Intelligence
ICT3909 (Final Year Project in Artificial Intelligence
– 30ECTS)
Proposal Form
Title:
Smart phone application for interactive line drawing of natural images
Project Supervisor:
Dr. George Azzopardi
Project Co-supervisor:
(if applicable)
Main Subject Area/s:
Smartphone applications, computer vision
The aim of this project is to improve the efficiency of the existing state-of-theart CORF algorithm and to implement it as a smartphone application. Just
like the image enhancement functionalities on smartphones we will add a new
app to extract the contours of a given picture. The extraction of contours may
also be performed in an interactive way. For instance, the user may be allowed
to extract the contours only of selected objects, such as faces. Also the user
Brief Project Description inc.
References:
(word limit approx. 300 words)
will be able to add, remove and modify strokes created by the algorithm.
The student will then evaluate the implemented app by comparing its improved
efficiency with other state of the art contour extraction algorithms and by
analyzing a user satisfaction questionnaire.
Downloads:
• Matlab code of the contour detection algorithm: http://tinyurl.com/kkkplnd
• Research paper of the CORF algorithm: http://tinyurl.com/pexz4xh
Resources Required:
Recommended Prerequisites /
Knowledge Required and
Supporting 3rd Year Study units:
Smart phone
No prerequisites
rd
Supporting 3 year study units: ICS3206, ICS3129
Foreseeable Ethical Issues and
How these will be tackled:
N/A
(if applicable)
Deadline: Monday 25th April 2016. To be submitted to Ms Francelle Scicluna
(francelle.scicluna@um.edu.mt)
Faculty of ICT
B.Sc. I.T. (Hons.) in Artificial Intelligence
ICT3909 (Final Year Project in Artificial Intelligence
– 30ECTS)
Proposal Form
Title:
Optical Music Recognition (OMR)
Project Supervisor:
Dr. George Azzopardi
Project Co-supervisor:
(if applicable)
Main Subject Area/s:
Dr. Alexandra Bonnici
Computer vision, image processing, music
Optical Music Recognition (OMR) is the automatic interpretation of printed
music sheets. It is an important application in the music domain with various
potential. It could for, for instance, be used to automatically convert a piece of
music to Braille, a music editor could make corrections to an old edition using a
music notation program, a virtual instructor could match the produced music of
a student with the one given in the sheet, among others.
In this project we will investigate computer vision approaches, such as the
COSFIRE algorithm, for the automatic interpretation of music sheets. Figure 1
shows an example of a piece of a music sheet with isolated notes. We will
evaluate the proposed system on real music sheets that are publicly available
or that we produce.
Brief Project Description inc.
References:
(word limit approx. 300 words)
Figure 1 - Example of piece of a music sheet with detected notes
This project will be in collaboration with the Systems and Control Engineering
Department of the Faculty of Engineering. In this project the student will learn
how to use a state-of-the-art methodology to propose a solution for a computer
vision application.
This is a continuation of work being carried out by the department of Systems &
Control, where music scores are already being converted to MIDI files for
playback and that the project seeks to extend the complexity of the music
scores that can be handled by this project.
Downloads:
• Matlab code of the COSFIRE algorithm: http://tinyurl.com/juy8rgo
• Research paper: http://www.cs.rug.nl/~george/articles/PAMI2013.pdf
Resources Required:
Recommended Prerequisites /
Knowledge Required and
Supporting 3rd Year Study units:
N/A
No prerequisites
rd
Supporting 3 year study units: ICS3206, ICS3129
Foreseeable Ethical Issues and
How these will be tackled:
(if applicable)
N/A
Faculty of ICT
B.Sc. I.T. (Hons.) in Artificial Intelligence
ICT3909 (Final Year Project in Artificial Intelligence
– 30ECTS)
Proposal Form
Title:
Object recognition from images
Project Supervisor:
Dr. George Azzopardi
Project Co-supervisor:
(if applicable)
Main Subject Area/s:
Dr. Reuben Farrugia
Computer vision, image classification
Object recognition is an effortless operation for human beings but it is very
challenging for computer systems. Object recognition is one of the major topics
in the field of computer vision. Examples of applications include the visual
system of a robot to navigate in a specific environment, surveillance cameras
for detection of persons and cars, automatic quality visual inspection which is
used mostly in the manufacturing industry, optical character recognition and
many others.
Brief Project Description inc.
References:
(word limit approx. 300 words)
The aim of this project is to pick a benchmark dataset that is available in
http://computervisiononline or any other data set selected by the student and
investigate different computer vision approaches to categorize images. In
particular, we will look at convolutional neural networks and the bag of visual
words approach.
References
• Convolutional neural network: http://tinyurl.com/oa2nlcd
• Bag of visual words: http://tinyurl.com/hk9v8fp
Resources Required:
Recommended Prerequisites /
Knowledge Required and
N/A
No prerequisites
rd
Supporting 3 Year Study units:
Supporting 3rd year study units: ICS3206, ICS3129
Foreseeable Ethical Issues and
How these will be tackled:
N/A
(if applicable)
Deadline: Monday 25th April 2016. To be submitted to Ms Francelle Scicluna
(francelle.scicluna@um.edu.mt)
Faculty of ICT
B.Sc. I.T. (Hons.) in Artificial Intelligence
ICT3909 (Final Year Project in Artificial Intelligence
– 30ECTS)
Proposal Form
Title:
Automatic Runway detection for Unmanned Aerial Vehicles (UAV)
Project Supervisor:
Dr. George Azzopardi
Project Co-supervisor:
(if applicable)
Main Subject Area/s:
Prof. David Zammit Mangion
Computer vision, image processing, machine learning, aviation
Unmanned aerial vehicles (UAVs), also known as drones, are commonly
preferred vehicles for dangerous scenarios. They function with different degree
of autonomy. In most of the cases they are controlled only by onboard
computers, making them fully autonomous.
One of the challenges of such vehicles is to automatically detect the runway of
an airport in order to align itself before landing.
The goal of this project is to use image processing and computer vision
techniques to detect the runway in real-time from aerial videos. It will be in
collaboration with the Institute of Aerospace Technologies who will provide us
Brief Project Description inc.
References:
(word limit approx. 300 words)
with the data and the domain knowledge. We will also use a simulator from
their institute. In particular, we will investigate techniques such the Hough
transform and the CORF algorithms to detect straight edges/lines that are
robust to noise.
Figure 1: (a) Example of a UAV and an (b) example of the detection of a runway
Downloads:
• Matlab code of CORF with Push-Pull inhibition: http://tinyurl.com/kkkplnd
• Research paper: http://tinyurl.com/pexz4xh
Resources Required:
N/A
Recommended Prerequisites /
Knowledge Required and
No prerequisites
rd
Supporting 3 Year Study units:
rd
Supporting 3 year study units: ICS3206, ICS3129
Foreseeable Ethical Issues and
How these will be tackled:
N/A
(if applicable)
Deadline: Monday 25th April 2016. To be submitted to Ms Francelle Scicluna
(francelle.scicluna@um.edu.mt)
Faculty of ICT
B.Sc. I.T. (Hons.) in Artificial Intelligence
ICT3909 (Final Year Project in Artificial Intelligence
– 30ECTS)
Proposal Form
Title:
Automatic Generation of Nonsense Words in Maltese
Project Supervisor:
Gordon Pace
Project Co-supervisor:
(if applicable)
Main Subject Area/s:
Claudia Borg
Natural Language Programming
This project will look at the automatic generation of nonsense words in Maltese
which sound as close to actual Maltese words as possible.
As a first step, a list of words and their phonetic transcriptions will be put
together as a dataset to (i) analyse mutually replaceable sounds in Maltese;
and (ii) train a classifier as to what Maltese sounds like.
Brief Project Description inc.
A program will then generate plausible words in Maltese, and test their
References:
closeness to actual Maltese words using a variety of metrics, as well as the
(word limit approx. 300 words)
classifier itself.
In order to evaluate the resulting nonsense word generator, we will be using
standard techniques similar to those used in psycholinguistic experiments,
primarily by having users react to the use of the words and assessing how
closely their reaction corresponds to real and random words.
Resources Required:
Recommended Prerequisites /
Knowledge Required and
Supporting 3rd Year Study units:
ICS2203, LIN3301
Faculty of ICT
B.Sc. I.T. (Hons.) in Artificial Intelligence
ICT3909 (Final Year Project in Artificial Intelligence
– 30ECTS)
Proposal Form
Title:
Project
Supervisor:
EXCITE – EXtracting geographiC Information from TExt
Dr Joel Azzopardi
Project Cosupervisor:
Mr Charlie Abela
(if applicable)
Main Subject
Knowledge Discovery, Text Processing, Web Intelligence
Area/s:
Geotagging is the process of adding geographic spatial information to an object (text,
multimedia, …). This geographic spatial information generally consists of longitude and
latitude coordinates. By this information, the object in question can be visualised on a map,
and distances can then be calculated accordingly.
Visualising knowledge spatially on a GIS interface such as Google Maps or Open Street Maps
can be very helpful for various reasons – e.g. to show what events are happening in your
vicinity, or checking where accidents have occurred.
Brief Project
This research aims to develop methods that can automatically identify references to locations
Description
in text, and uses available spatial data, such as that available from openstreetmap.org to
inc.
geotag that text. Through this service, one can then geo-tag news articles and visualise them
References:
on a GIS map. This spatial information can be enhanced using information from other
(word limit
sources – such as the police notices from the Department of Information detailing road
approx. 300
closures
words)
(https://www.gov.mt/en/Government/Government%20Gazette/Police%20Notices/Pages/Road
-closures-in-Malta-and-Gozo.aspx). In this way, users will be presented with a better view of
what is happening in their locations of interest.
The main research questions tackled by this research include:
•
How can geographic information be extracted from unstructured text, and converted
into numeric coordinates?
•
To what level of detail can the extracted geographic information be resolved?
•
How can geotagging information be used to enhance the presentation of information
to users?
Resources
Personal Computer
Required:
Recommende
This research builds upon material covered in:
d
•
Prerequisites /
ICS2205 – Web Intelligence
Knowledge
Required and
rd
Recommended 3 Year Study Units:
rd
Supporting 3
•
ICS3208 – Advanced Web Intelligence
Year Study -
•
ICS3207 – Knowledge Discovery and Management
units:
Foreseeable
Ethical Issues
and How these
N/A
will be tackled:
(if applicable)
Deadline: Monday 25th April 2016. To be submitted to Ms Francelle Scicluna
(francelle.scicluna@um.edu.mt)
Faculty of ICT
B.Sc. I.T. (Hons.) in Artificial Intelligence
ICT3909 (Final Year Project in Artificial Intelligence
– 30ECTS)
Proposal Form
Title:
VANITAS – Visualising and AugmeNting Interesting Text collectionS
Project Supervisor:
Dr Joel Azzopardi
Project Co-supervisor:
(if applicable)
Main Subject Area/s:
Mr Charlie Abela
Text Mining, Query Generation, Information Retrieval
In 2015, a scandal erupted in 2015 when it became known that Hillary
Clinton was using non-governmental servers for emails during the time when
she was serving as secretary of state. Following legal issues and freedom
of information requests, the USA state department released Hillary Clinton's
emails to the public.
Whilst many people would like to analyse such an interesting document
collection, very few would actually read through all the documents. The aim
of this project is to research and develop methods whereby users can obtain
an overview, and perform analyses of interesting document collections
(such as the Hillary Clinton Email set) without needing to read the whole
Brief Project Description inc.
text. Solutions are expected to include the use of web visualisation libraries
References:
(such as d3.js), and allow users to 'drill-down' in the information.
(word limit approx. 300 words)
Certain data collections can be rendered more useful for users by linking the
information there with data from other sources. In the case of the Hillary
Clinton email collections, users would definitely find it more useful, if the
information there can be placed within the context of the world events at that
time. In view of this, a second aim of this project is to identify relationships
from datasets to corresponding news events.
The research questions tackled by this research are:
•
How can text collections be visualised in order to provide overview
of their content to users?
•
How can the context of information in a collection be augmented by
linking it to corresponding news reports?
Resources Required:
Personal Computer
This research builds upon material covered in:
Recommended Prerequisites /
- ICS2205 – Web Intelligence
Knowledge Required and
- ICS2208 – Intelligent Interfaces I
rd
Supporting 3 Year Study rd
units:
Recommended 3 Year Study Units:
-ICS3208 – Advanced Web Intelligence
Foreseeable Ethical Issues and
How these will be tackled:
N/A
(if applicable)
Deadline: Monday 25th April 2016. To be submitted to Ms Francelle Scicluna
(francelle.scicluna@um.edu.mt)
Faculty of ICT
B.Sc. I.T. (Hons.) in Artificial Intelligence
ICT3909 (Final Year Project in Artificial Intelligence
– 30ECTS)
Proposal Form
Title:
What's up? Research
Project Supervisor:
Dr. Joel Azzopardi
Project Co-supervisor:
(if applicable)
Main Subject Area/s:
Dr. Charlie Abela
Text Mining, Information Extraction
One of the problems faced by researchers is identifying conferences,
workshops and other events of interest to them and that are happening in
the near future. A researcher's interest in such events depends mostly on
whether the event's theme (and topics) is in line with the researcher's areas
of interest. Moreover, the researcher will not be potentially interested in an
event for the sake of attending to that event, but may also be interested in
submitting one or more publications and/or participate in a challenge that
may be organised during that event.
The aim of this project is to develop a system that is able to search through
Brief Project Description inc.
References:
(word limit approx. 300 words)
suitable portals (e.g. DBWorld) and automatically identify conferences
and/or workshops of interest for different researchers, and extract the
relevant information.
The same event may be referred to in different
sources, and the information supplied might be different, thus the ideal
system should be able to aggregate the data about the same event available
from different sources. The output information should be structured such
that important dates for the events of interest may be automatically inserted
into a user's calendar.
This project will involve research on how one can automatically extract
information about conferences and workshops from unstructured and semistructured sources, possibly with the help of existing tools. The extracted
details associated with each event may include: important dates, themes,
impact, etc.
Research questions tackled by this research include:
- How can information on research events be mined successfully from
unstructured or semi-structured sources?
- How can information from different sources be fused together in order to
minimise repetition but ensure the presentation of all the available
information?
Resources Required:
Personal Computer
This research builds upon material covered in:
Recommended Prerequisites /
•
ICS2205 – Web Intelligence
•
ICS2203 – Natural Language Processing: Methods and Tools
Knowledge Required and
Supporting 3rd Year Study -
rd
Recommended 3 Year Study Units:
units:
•
ICS3208 – Advanced Web Intelligence
•
ICS3207 – Knowledge Discovery and Management
Foreseeable Ethical Issues and
How these will be tackled:
N/A
(if applicable)
Deadline: Monday 25th April 2016. To be submitted to Ms Francelle Scicluna
(francelle.scicluna@um.edu.mt)
Faculty of ICT
B.Sc. I.T. (Hons.) in Artificial Intelligence
ICT3909 (Final Year Project in Artificial Intelligence
– 30ECTS)
Proposal Form
Title:
Question Answering using Wikipedia
Project Supervisor:
Dr Joel Azzopardi
Project Co-supervisor:
(if applicable)
Main Subject Area/s:
Dr Chris Staff
Text Mining, Information Retrieval
One of the projects carried out by the Allen Institute for AI (a foundation
aiming to develop human-level intelligence) is the ARISTO project – a
system that acquires and stores a vast amount of knowledge in computable
form, then applies this knowledge to answer a variety of science questions
from standardized exams for students in multiple grade levels
(http://allenai.org/aristo/).
Such research falls under the domain of Question Answering whereby
systems attempt to automatically answer questions that are posed to them in
natural language. Question Answering Systems can utilise various types of
Brief Project Description inc.
References:
(word limit approx. 300 words)
sources – from unstructured text collections, to structured sources such as
databases, and even search engines.
The main aim of this FYP is to perform Question Answering using only the
English language Wikipedia as the information source. Using Wikipedia
provides various advantages such as: its wide coverage; the availability of
the data for free; and the availability of a search API that has no usage
constraints.
Questions can range from trivial questions such as “When was Charles de
Gaulle born?” to hard questions such as “What trail did Lincoln use a
Farmers' Almanac in?”. One of the subsidiary aims of this research would
be to assess to what level of question difficulty that the developed system
can handle.
Resources Required:
Personal Computer
This research builds upon material covered in:
Recommended Prerequisites /
•
ICS2205 – Web Intelligence
•
ICS2203 – Natural Language Processing: Methods and Tools
Knowledge Required and
Supporting 3rd Year Study units:
Recommended 3rd Year Study Units:
•
ICS3208 – Advanced Web Intelligence
•
ICS3207 – Knowledge Discovery and Management
Foreseeable Ethical Issues and
How these will be tackled:
N/A
(if applicable)
Deadline: Monday 25th April 2016. To be submitted to Ms Francelle Scicluna
(francelle.scicluna@um.edu.mt)
Faculty of ICT
B.Sc. I.T. (Hons.) in Artificial Intelligence
ICT3909 (Final Year Project in Artificial Intelligence
– 30ECTS)
Proposal Form
Title:
Intelligent Bookmark Organiser and Recommender
Project Supervisor:
Dr Joel Azzopardi
Project Co-supervisor:
(if applicable)
Dr Colin Layfield
Document Categorisation, Information Extraction, Latent Semantic Analysis,
Main Subject Area/s:
Information Recommendation
The aim of this research project is to explore the possibility of organising a
user's browser bookmarks according to his/her browsing history. This
project will ultimately involve the implementation of a browser extension that
can organise a user's bookmarks into a set of categories or hierarchies.
This process can be done in a fully unsupervised manner, or in a semisupervised manner using an initial hierarchy that was done by the user.
Latent Semantic Analysis (LSA) can be used to help in this clustering. The
systems will furthermore propose further pages to be bookmarked according
Brief Project Description inc.
References:
(word limit approx. 300 words)
to the user's browsing history. A possible addition to this project can be that
of performing some minor extractive summarisation whereby a short
summary of the recommended/bookmarked pages is constructed (through
sentence/phrase extraction) and made available to the user.
The main research questions tackled include:
•
Can a high-quality hierarchy of bookmarks be constructed
automatically with the help of LSA?
•
Can the user browsing history be 'mined' to identify pages that prove
useful on the user's bookmark list.
Resources Required:
Personal Computer
This course builds upon material covered in:
Recommended Prerequisites /
- ICS2205 – Web Intelligence
Knowledge Required and
- ICS2208 – Intelligent Interfaces I
rd
Supporting 3 Year Study Recommended 3rd Year Study Units:
units:
ICS3208 – Advanced Web Intelligence
Foreseeable Ethical Issues
This system will processes the user's browsing data. However, it will work
and How these will be tackled:
entirely from the client's side – i.e. no information is shared beyond the
(if applicable)
user's computer.
Deadline: Monday 25th April 2016. To be submitted to Ms Francelle Scicluna
(francelle.scicluna@um.edu.mt)
Faculty of ICT
B.Sc. I.T. (Hons.) in Artificial Intelligence
ICT3909 (Final Year Project in Artificial Intelligence
– 30ECTS)
Proposal Form
Title:
Project Supervisor:
A domain-specific language to implement, test, and visualise grammatical
inference algorithms
Kristian Guillaumier
Project Co-supervisor:
(if applicable)
Grammatical inference, machine learning, programming languages,
Main Subject Area/s:
visualization
Designing grammatical inference algorithms using general purpose
programming languages requires the implementation of a lot of boiler-plate
code to manage complex data structures and large data sets. Also, due to the
nature of the problem, visualizing and understanding the behavior of these
algorithms is hard.
The aim of this FYP is to design and implement a high-level, domain-specific
Brief Project Description inc.
programming language with the following attributes:
•
References:
Easy creation and manipulation of finite state machines. This will
involve creating several specialized data types and operators.
(word limit approx. 300 words)
•
Ensure that the language is rich enough to implement several GI
algorithms.
•
A visual debugging interface to allow users to step through, trace, and
visualize (e.g. the intermediate state of the FSM) the behavior of the
algorithms implemented.
In this FYP we will focus on the class of regular languages.
Resources Required:
n/a
Recommended Prerequisites /
Knowledge Required and
rd
Supporting 3 Year Study -
Formal languages and automata, comiling techniques, machine learning.
units:
Foreseeable Ethical Issues and
How these will be tackled:
n/a
(if applicable)
Deadline: Monday 25th April 2016. To be submitted to Ms Francelle Scicluna
(francelle.scicluna@um.edu.mt)
Faculty of ICT
B.Sc. I.T. (Hons.) in Artificial Intelligence
ICT3909 (Final Year Project in Artificial Intelligence
– 30ECTS)
Proposal Form
Title:
Using automatic greyscale image colourisation as a compression technique
Project Supervisor:
Kristian Guillaumier
Project Co-supervisor:
(if applicable)
Main Subject Area/s:
Machine learning, image processing, data compression/signal processing
Lossy image compression algorithms exhibit several artifacts in the resulting
output including ringing, posterizing, blocking, colour shifts, and blurring. At low
bitrates the intensity of these artifacts is exacerbated and can be easily
detected by viewers.
This FYP is based on the following observations:
•
The human eye is not as sensitive to colour as it is to brightness (the
premise of chroma subsampling).
•
Algorithms exist that use ‘intelligent’ flood-fill methods to colourise
greyscale images accurately and reliably.
•
Greyscale images require less data to store than colour images.
Brief Project Description inc.
References:
(word limit approx. 300 words)
In this FYP we will design, build, and evaluate a codec that:
•
Implements a lossy compression algorithm to a luminance-only version
of the image. The hypothesis is that higher bitrates can be used to
obtain equivalent file sizes to compressing colour images at lower
bitrates.
•
This would result in a compressed greyscale image that has fewer
artifacts at the same file size of a compressed colour image.
•
Search a space of ‘flood-fill points’ finding an arrangement that when
applied to colourise the compressed greyscale image would result in a
colour image that most closely matches the raw input.
•
The optimal colour point set is simply a list of image coordinates and
colour values that should be negligible in size.
•
The expectation is that keeping the output file size constant, many
artifacts including ringing, blocking, and blurring can be reduced at the
expense of colour fidelity when compared to image compression
techniques such as JPEG. Of course, the assumption is that people
are less sensitive to loss in colour quality compared to, say, sharpness
– this is especially true if they possess no a priori knowledge about the
colours in the original uncompressed image.
Resources Required:
n/a
Recommended Prerequisites /
Knowledge Required and
Supporting 3rd Year Study -
Search and optimization, signal processing, machine learning.
units:
Foreseeable Ethical Issues and
How these will be tackled:
n/a
(if applicable)
Deadline: Monday 25th April 2016. To be submitted to Ms Francelle Scicluna
(francelle.scicluna@um.edu.mt)
Faculty of ICT
B.Sc. I.T. (Hons.) in Artificial Intelligence
ICT3909 (Final Year Project in Artificial Intelligence
– 30ECTS)
Proposal Form
Title:
Optimizing Light Path Generation via a Metaheuristic Approach
Project Supervisor:
Keith Bugeja
Project Co-supervisor:
(if applicable)
Main Subject Area/s:
Sandro Spina
Metaheuristics, Search and Optimization
The realistic rendering of virtual environment requires the faithful simulation of
light exchange between surfaces. The path tracing algorithm is a Monte Carlo
solution for image synthesis which simulates photon interactions between
various types of surfaces. The algorithm can yield a coarse solution or, given
more time, converge towards a more accurate one. The convergence rate is
highly dependent upon the type of surface materials present in the
Brief Project Description inc.
References:
(word limit approx. 300 words)
environment, with some solutions taking considerably longer due to
phenomena such as caustics or geometric slits. Path tracing and other similar
stochastic methods use random walks to estimate radiant flux in an
environment.
The aim of this project is to use probabilistic techniques to find suitable light
paths through a scene. In particular, metaheuristic optimizations will be
employed to improve light path generation for synthesized images that present
high solution variance. It is hypothesized that the improved paths will reduce
solution convergence time, thus accelerating the rendering process.
Resources Required:
Recommended Prerequisites /
Knowledge Required and
rd
Supporting 3 Year Study units:
C/C++
Foreseeable Ethical Issues and
How these will be tackled:
(if applicable)
Deadline: Monday 25th April 2016. To be submitted to Ms Francelle Scicluna
(francelle.scicluna@um.edu.mt)
Faculty of ICT
B.Sc. I.T. (Hons.) in Artificial Intelligence
ICT3909 (Final Year Project in Artificial Intelligence
– 30ECTS)
Proposal Form
Title:
Project Supervisor:
Digital footprint analytics for prediction, detection and management of
neurological disorders
Dr Lalit Garg
Project Co-supervisor:
(if applicable)
It is an application of AI techniques (especially machine learning) to social
media data analytics.
Main Subject Area/s:
Machine learning, digital footprints, data analytics, social media analytics,
behavioral health, mental health, neurological disorders, sentiment analysis,
opinion mining, natural language processing
A large population is affected by neurological disorders such as depression,
epilepsy, stroke, parkinsion and other mental health issues. Early detection
might help in better treatment of these psychiatric symptoms. As we are getting
more and more active online. The data generated through one’s online
activities (called digital footprint) might provide some clue about his/her mental
health status. In this project we will investigate if we can use digital footprint to
assess one’s mental health status and to timely predict, detect and manage
his/her psychiatric symptoms.
1.
2.
Brief Project Description inc.
References:
3.
(word limit approx. 300 words)
4.
5.
Libert, T., Grande, D. and Asch, D.A., 2015. What web browsing reveals about your
health. BMJ, 351, p.h5974.
Bechmann, A. and Vahlstrup, P.B., 2015. Studying Facebook and Instagram data:
The Digital Footprints software. First Monday, 20(12).
Oatley, G., Crick, T. and Mostafa, M., 2015. Digital Footprints: Envisaging and
Analysing Online Behaviour. In Proceedings of 2015 Symposium on Social Aspects
of Cognition and Computing Symposium (SSAISB 2015).
Youyou, W., Kosinski, M. and Stillwell, D., 2015. Computer-based personality
judgments are more accurate than those made by humans.Proceedings of the
National Academy of Sciences, 112(4), pp.1036-1040.
Asch, D.A., Rader, D.J. and Merchant, R.M., 2015. Mining the social
mediome. Trends in molecular medicine, 21(9), pp.528-529.
6.
Feher,
7.
Psychology, 10, p.9789814723398_0007.
Zhang, D., Guo, B., Li, B. and Yu, Z., 2010. Extracting social and community
intelligence from digital footprints: an emerging research area. In Ubiquitous
8.
9.
K.,
2016. Digital
identity: The transparency
of the self. Applied
Intelligence and Computing (pp. 4-18). Springer Berlin Heidelberg.
Gencoglu, O., Simila, H., Honko, H. and Isomursu, M., 2015, August. Collecting a
citizen's digital footprint for health data mining. In Engineering in Medicine and
Biology Society (EMBC), 2015 37th Annual International Conference of the
IEEE (pp. 7626-7629). IEEE.
Mulder, D. J. (2015). Social Media: Digital Footprints and Digital Wisdom. Faculty
Work: Comprehensive List. Paper 424.
10. Marchal, A., Lejeune, P. and Nico, D.B., 2015. Blueprints for unique footprints: sex,
age and individual identification from digital 3D models of lion (Panthera leo) paws.
13th Savanna Science Network Meeting, Skukuza, 9-12 March 2015.
Internet Access, Matlab, MS-Visual Studio, Library access, Access to
high quality literature resources, tools for sentiments analysis,
Resources Required:
natural language processing tools, access to digital footprints of
participants.
Knowledge of basic AI methods, good programming skills in a language of your
choice, strong analytical and problem solving skills, fast learning abilities,
Recommended Prerequisites /
Knowledge Required and
Supporting 3rd Year Study units:
Foreseeable Ethical Issues and
How these will be tackled:
(if applicable)
reliable, responsible, hardworking, enthusiasm and determination to learn and
acquire new skills.
B.Sc. IT (Hons)- Artificial Intelligence study units:
ICS3208
Advanced Web Intelligence
ICS3207
Knowledge Discovery and Management
CSA3206
Machine Learning, Expert Systems and Fuzzy Logic
LIN3012
Data Driven Natural Language Processing
Students would analyse the health status and social media data of users. This
might raise privacy issues. A written consent will be obtained from participants
before the start of the project. Also, a prior approval will be obtained from the
University Research Ethics Committee.
Deadline: Monday 25th April 2016. To be submitted to Ms Francelle Scicluna
(francelle.scicluna@um.edu.mt)
Faculty of ICT
B.Sc. I.T. (Hons.) in Artificial Intelligence
ICT3909 (Final Year Project in Artificial Intelligence
– 30ECTS)
Proposal Form
Title:
Project Supervisor:
Project Co-supervisor:
(if applicable)
Hospital admission pattern analysis, bed resource requirements forecasting,
allocation and management
Dr Lalit Garg
Mr Kristian Guillaumier
The project would develop novel applications of AI/ML methods in the
healthcare management problems such as hospital admission pattern analysis,
bed resource requirements forecasting, allocation and management. It would
Main Subject Area/s:
provide students an excellent opportunity to understand how AI/ ML methods
can be used for developing solutions to real life problems and also assessing
effectiveness of such AI/ML tools.
Healthcare resource planners need to develop policies that ensure optimal
allocation of scarce healthcare resources. This goal can be achieved by
analysing admission patterns to forecast daily resource requirements to ensure
optimum allocation and management of available resources. If resources are
limited, admission should be scheduled according to the resource availability.
Such resource availability or demand can change with time. We here model
admissions and patient flow through the care system as a discrete time Markov
chain. In order to have a more realistic representation, a non-homogeneous
model is developed which incorporates time-dependent covariates, namely a
Brief Project Description inc.
References:
(word limit approx. 300 words)
patient’s present age and the present calendar year. However, more
sophisticated models are required to better manage changes in admission
patterns and resource requirements. As our previous work, we have already
developed many such sophisticated models for better modelling admission
patterns and resource requirements [1-4]. In this final year project (FYP), we
will extend our work to develop novel approaches to effectively solve the
problem using AI/ML based methods.
1. Garg L, McClean SI, Meenan BJ, Millard PH (2010). A nonhomogeneous discrete time Markov model for admission scheduling
and resource planning in a care system. Health Care Management
Science. 13(2):155–169.
2. Garg L, McClean SI, Meenan BJ, Millard PH (2009). Nonhomogeneous Markov Models for Sequential Pattern Mining of
Healthcare Data. IMA journal Management Mathematics. 20(4): 327344.
3. Garg L, McClean SI, Meenan BJ, Barton M, Fullerton K (2012).
Intelligent patient management and resource planning for complex,
heterogeneous and stochastic healthcare systems. In press. IEEE
Transactions on Systems, Man, and Cybernetics--Part A: Systems and
Humans.
4. Garg L, McClean SI, Meenan BJ, Barton M, Fullerton K (2013). An
Extended Mixture Distribution Survival Tree for Patient Pathway
Prognostication. Communications in Statistics: Theory and
Methodology. 42(16):2912-2934.
Resources Required:
A computing device (PC or Laptop), Internet Access, Matlab, C/C++, UM
Library access, Access to high quality literature resources.
Knowledge of basic AI methods, good programming skills in a language of your
choice, strong analytical and problem solving skills, fast learning abilities,
reliable, responsible, hardworking, enthusiasm and determination to learn and
Recommended Prerequisites /
acquire new skills.
Knowledge Required and
B.Sc. IT (Hons)- Artificial Intelligence study units:
rd
Supporting 3 Year Study -
ICS3207
Knowledge Discovery and Management
units:
CCE3502
Modelling and Computer Simulation
ICS3206
Machine Learning, Expert Systems and Fuzzy Logic
CPS3239
Computability and Complexity
(Preferable)
Foreseeable Ethical Issues and
We have already got ethical approval for the project. Students would also
How these will be tackled:
require to apply for an ethical approval to the University’s research ethics
(if applicable)
committee.
Deadline: Monday 25th April 2016. To be submitted to Ms Francelle Scicluna
(francelle.scicluna@um.edu.mt)
Faculty of ICT
B.Sc. I.T. (Hons.) in Artificial Intelligence
ICT3001 (Final Year Project in Artificial Intelligence)
Proposal Form
Title:
Machine learning methods for handling missing data in medical questionnaires
Project Supervisor:
Dr Lalit Garg
Project Co-supervisor:
(if applicable)
Mr Kristian Guillaumier
The project would assess and develop novel applications of AI/ML methods in
missing data handling in medical questionnaires. It would provide students an
Main Subject Area/s:
excellent opportunity to understand how AI/ ML methods can be used for real
life problems and also developing skills in assessing and developing AI/ML
tools.
The proposed project is a part of a University funded project. Self-report
questionnaires are used as an extremely valuable instrument to assess the
quality of life of a patient, its relationship with socioeconomic and
environmental factors, disease risk/ progress, treatment and disease burden,
treatment response and quality of care. However, a common problem with such
questionnaires is missing data. Despite enormous care and effort to prevent it,
some level of missing data is common and unavoidable. Such missing data can
have a detrimental impact on statistical analyses based on the questionnaire
responses. A variety of methods have been suggested for missing data
imputation. Nevertheless, more research is desperately needed to assess and
Brief Project Description inc.
References:
(word limit approx. 300 words)
improve the reliability of data imputation. We have already developed/
proposed some novel methods to handle missing data [1-4]. In this final year
project (FYP), we will extend our work to develop novel machine learning
procedures enforcing collaborative filtering to complete missing data in medical
questionnaires. Empirical datasets would be used to assess these novel
methods against existing methods.
1. Garg L, Dauwels J, Earnest A, Pang L (2013) Tensor based methods
for handling missing data in quality-of-life questionnaires. IEEE Journal
of Biomedical and Health Informatics. In press. doi:
10.1109/JBHI.2013.2288803.
2. Asif MT, Srinivasan K, Garg L, Dauwels J, Jaillet P (2013) Lowdimensional Models for Missing Data Imputation in Road Networks,
ICASSP 2013, May 26 - 31, 2013.
3. Dauwels J, Garg L, Earnest A, Pang LK (2012). Tensor Factorizations
for Missing Data Imputation in Medical Questionnaires, The 37th
International Conference on Acoustics, Speech, and Signal Processing
(ICASSP), Kyoto, Japan, March 25 - 30, 2012.
4. Dauwels J, Garg L, Earnest A, Pang LK (2011). Handling Missing Data
in Medical Questionnaires Using Tensor Decompositions. The Eighth
International Conference on Information, Communications, and Signal
Processing (ICICS 2011). Singapore 13-16 December, 2011.
Resources Required:
A computing device (PC or Laptop), Internet Access, Matlab, C/C++, UM
(if student needs to buy them)
Library access, Access to high quality literature resources.
Knowledge of basic AI methods, good programming skills in a language of your
choice, strong analytical and problem solving skills, fast learning abilities,
Recommended Prerequisites /
Knowledge Required and
rd
Supporting 3 Year Studyunits:
reliable, responsible, hardworking, enthusiasm and determination to learn and
acquire new skills.
B.Sc. IT (Hons)- Artificial Intelligence study units:
CPS3239
Computability and Complexity
ICS3207
Knowledge Discovery and Management
ICS3206
Machine Learning, Expert Systems and Fuzzy Logic
(Preferable)
Foreseeable Ethical Issues and
We have already got ethical approval for the project. Students would also
How these will be tackled:
require to apply for an ethical approval to the University’s research ethics
(if applicable)
committee.
Deadline: Friday 17th of April 2015. To be submitted to Ms FrancelleScicluna
(francelle.scicluna@um.edu.mt)
Faculty of ICT
B.Sc. I.T. (Hons.) in Artificial Intelligence
ICT3909 (Final Year Project in Artificial Intelligence
– 30ECTS)
Proposal Form
Title:
Smart Sensor for EEG Acquisition and Epileptic Seizure Detection
Project Supervisor:
Dr Lalit Garg
Project Co-supervisor:
(if applicable)
Mr Kristian Guillaumier
The project would assess and develop novel applications of AI/ML methods in
EEG analysis and epileptic seizure detection. It would provide students an
Main Subject Area/s:
excellent opportunity to understand how AI/ ML methods can be used for real
life problems and also developing skills in assessing and developing AI/ML
tools.
Epilepsy is a serious neurological disease that has an adverse socio-economic
impact on a substantial segment of the world population. A number of studies
have been carried out in the past to explore the feasibility of a practical realtime seizure detector for effective management and mitigation of this disease
[1]. With this goal in sight we have already developed/ proposed some novel
methods using machine learning techniques such as singular vector machine
(SVM) and extreme learning machine (ELM) [2-3] for seizure detection using
an exhaustive scalp EEG data collected from pediatric patients. [2]. In this final
year project (FYP), we will extend our work to develop novel machine learning
Brief Project Description inc.
approaches for seizure detection. Empirical datasets would be used to assess
References:
these novel approaches against existing approaches.
(word limit approx. 300 words)
1. Ali H. Shoeb, John V. Guttag: Application of Machine Learning To
Epileptic Seizure Detection. ICML 2010: 975-982.
2. Huang, G. B., Wang, D. H., & Lan, Y. (2011). Extreme learning
machines: a survey. International Journal of Machine Learning and
Cybernetics, 2(2), 107-122.
3. Huang, G. B., Zhu, Q. Y., & Siew, C. K. (2006). Extreme learning
machine: theory and applications. Neurocomputing, 70(1), 489-501.
4. Agrawal A, Garg L, Dauwels J (2013) Application of empirical mode
decomposition algorithm for epileptic seizure detection from scalp EEG,
The 35th Annual International Conference of the IEEE Engineering in
Medicine and Biology Society (EMBC’13) Osaka, Japan, 3-7 July 2013.
Resources Required:
A computing device (PC or Laptop), Internet Access, Matlab, C/C++, UM
Library access, Access to high quality literature resources.
Knowledge of basic AI methods, good programming skills in a language of your
choice, strong analytical and problem solving skills, fast learning abilities,
Recommended Prerequisites /
Knowledge Required and
Supporting 3rd Year Study units:
reliable, responsible, hardworking, enthusiasm and determination to learn and
acquire new skills.
B.Sc. IT (Hons)- Artificial Intelligence study units:
CCE3206
Digital Signal Processing
ICS3206
Machine Learning, Expert Systems and Fuzzy Logic
ICS3207
Knowledge Discovery and Management
CPS3239
Computability and Complexity (preferable)
Foreseeable Ethical Issues and
We have already got ethical approval for the project. Students would also
How these will be tackled:
require to apply for an ethical approval to the University’s research ethics
(if applicable)
committee.
Deadline: Monday 25th April 2016. To be submitted to Ms Francelle Scicluna
(francelle.scicluna@um.edu.mt)