Master Thesis

Sujets de mémoires/ Master thesis subject MLG 2012-2013

Le Machine Learning Group propose pour l'année 2012/2013 une dizaine de sujets pour les étudiants en master. Les domaines d'applications incluent la bioinformatique, les réseaux de capteurs, l'évolution artificielle, la médecine assistée par ordinateur, les protéines artificielles et la dynamique des réseaux. 

NB: Le nombre de sujets est limité. Les étudiants intéressés sont priés de se manifester au plus tôt.

Topics:

1. Machine learning for motion analysis in the treatment of spasticity (Yann-Aël Le Borgne, Gianluca Bontempi)

2. Computational Biology: Investigations into protein structure and function (Tom Lenaerts, Elisa Cilia)

3. Computational Biology:Evolutionary Dynamics of advanced strategies  (Tom Lenaerts)

4. Computational Biology: Stochastic dynamics of chronic myeloid leukemia (Tom Lenaerts)

5. L'apprentissage automatique pour vérifier la sécurité de devices cryptographiques (TFE encadré par QualSec et MLG (machine learning group))

6. Machine learning for wireless sensor networks (Yann-Aël Le Borgne, Gianluca Bontempi)

7. Trading challenge (Gianluca Bontempi, Andrea Dal Pozzolo)

8. Sentiment analysis on Facebook (Yann-Aël Le Borgne, Gianluca Bontempi)

1. Machine learning for motion analysis in the treatment of spasticity (Yann-Aël Le Borgne, Gianluca Bontempi)

The treatment of muscle spasticity can nowadays take advantage of a large amount of clinical data. This is made possible by the development of new sensor technologies (e.g. camera, magnetic sensors, wearable inertial sensors like gyroscopes, accelerometers) and their integration in daily life monitoring systems. This opens the way to the development of data-driven approaches to modelling and detection of human movement with the purpose of obtaining better diagnosis of patients (e.g. affected by Parkinson's disease), improving the medication process and recognizing the movement patterns (e.g. in biometrics). The Master thesis will focus on automated approaches based on statistical machine learning and data mining approaches to emulate the innate human capability of recognizing, disambiguating, classifying the type of movements and to support clinicians in diagnosing and decision making.

The thesis will be carried out in the context of the ICT4REHAB project funded by the Brussels Region.

Required skills: statistical analysis, numerical computing, machine learning, passion for interdisciplinary research

2. Computational Biology : Investigations into protein structure and function (Tom Lenaerts, Elisa Cilia)

One of the main research tracks in the group is linked to questions related to the structure and function of proteins. Machine learning methods can assist in answering these questions.This  is a short list of topics which we want to investigate:

1. Network analysis of protein structures: Proteins are amino acid sequences that fold into a three-dimensional structure. To understand their structure one could replace every amino acid by a node and connect these nodes into a graph or network, where links define the proximity between the nodes. The aim of this project is to examine these graph features of proteins in order to improve our understanding of how proteins process information.

2. Creating artificial proteins from sequence and structural data: Once you know which residues in a protein sequence are important for their function, can you produce an artificial protein sequence that folds into the same structure and has a function equivalent to the original one? Recently it was shown that this actually may work. The aim of this project is to investigate the algorithm that was developed to solve this problem, on the one hand, and, on the other hand, to develop other one using structural information produced by an algorithm designed by the promoter of this project. 

3. Mining relevant features that drive protein function : Understanding how proteins behave is a complex task. Mutants of the same protein provide an invaluable source of data for the application of statistical learning techniques. Mutation data can be analyzed to find patterns of amino acids that are most relevant for the manifestation of a certain protein behavior. This thesis proposal aims at learning relevant features and rules to explain a specific protein behavior. The techniques used for attaining this goal will draw on game theory (used for feature selection) and/or inductive logic programming.

4. Preference handling as an approach to analyze and understande protein binding preferences: Proteins and especialy their domains have a finely tuned preference for particular peptides.  One of the main interests of bioinformaticians is to provide a description of these preferences so that potential binding partners can be searched with the database of known proteins.  This project proposes to investigate the relevance of preference handling methods to solve this problem. 

Interested? Contact Tom Lenaerts or Elisa Cilia for more details.

3. Computational Biology : Evolutionary Dynamics of Advanced Strategies (Tom Lenaerts)

Many strategic situations are characterized by player’s preferences that take into account the well being of others.  These preferences guide humans in their choices with whom to participate in economical or social activities.  Little attention has been given to how group formation shapes players’ beliefs concerning preferences and how preferences guide the formation of groups. We investigate how these other regarding preferences influence the matching of players, both in pairwise and N-player scenarios, and vice versa how group dynamics shape the beliefs concerning preferences (i.e. formation and adaptation of belief models) of the individual agent’s participating in concrete strategic problems.  

4. Computational Biology: Stochastic dynamics of chronic myeloid leukemia  (Tom Lenaerts)

The aim of this project is to investigate, through a model of the hematopoietic system and CML, the emergence and dynamics of therapy resistant clones, and the relation between patient treatment response, survival and the diagnostic risk groups.  Patients diagnosed with early-phase CML may relapse during treatment due to the appearance of cancer cells resistant to first-line treatment compounds like Imatinib. Understanding therefore how treatment affects the dynamics of these resistant cells is important and resulting insights will aid medical practitioners in setting up treatment protocols for individual patients.   In addition, each patient responds differently to Imatinib. Using our model and available serial Q-RT-PCR patient data we can determine the severity of the disease and the quality of initial treatment response.  Together these will reclassify patients with respect to their survival chances. Additionally it will shed light on the correlation between the risk groups identified at diagnosis and treatment response, which is not clear yet.  

Interested? Contact Tom Lenaerts for more details.

Required skills: Programming skills and passion for interdisciplinary research

Course prerequisite: INFO-F-305 (Modélisation et Simulation) or some equivalent course.

5. L'apprentissage automatique pour vérifier la sécurité de devices cryptographiques (TFE encadré par QualSec et MLG (machine learning group))

Description: Les devices cryptographiques sont utilisés dans de nombreux domaines (militaires, bancaires, automobiles, ...), ils permettent d'assurer entre autre la confidentialité, l'authenticité et l'intégrité des données. Ces devices doivent résister aux attaques cryptanalytiques connues. Une des techniques cryptanalytiques très efficaces est celle des attaques par canaux auxiliaires (side channel attacks). Celles-ci se focalisent sur le comportement des devices physiques (la consommation d'énergie ou le temps de calcul, ...) pour vérifier si une information secrète (e.g. la clé cryptographique) peut en être déduite. Toutefois, ces techniques peuvent être améliorées en les combinant avec celles de l'apprentissage automatique. Le travail se focalisera sur cette combinaison novatrice dans le but de tester des devices cryptographique dans des situations réelles, au sein de l'entreprise Atos Worldline spécialisée dans le milieu bancaire, et/ou au sein du concours DPA Contest (http://www.dpacontest.org).

Contact : Olivier Markowitch et Gianluca Bontempi

6. Machine learning for wireless sensor networks (Yann-Aël Le Borgne and Gianluca Bontempi)

Wireless sensor networks (WSN) form a new class of computing devices, able to monitor our environment with a high spatiotemporal resolution. These networks are composed of tens or hundreds of tiny computing devices, called motes, able to sense, process, and communicate data. A typical mote has a volume of a few millimeter/centimeter cube, sensors such as light, temperature, or sound, and a microcontroller of a few MHz. The wireless radio has a range of tens of meters, and a few hundreds of kbps. Applications for these networks are numerous, in domains such as ecology, medicine, industry, agriculture, or defense. The thesis will consist in designing machine learning techniques for improving the usability and the efficiency of WSN. The application considered will be environmental monitoring, where a set of motes periodically collect data (such as temperature), and send them to a base station (a high-end computer), where the spatiotemporal evolution of the measurements can be observed. In these applications, data is often strongly correlated, and it is possible to significantly reduce the traffic in the network by using machine learning techniques which predict or compress the data as they flow in the network. The practical benefits are to reduce the energy consumption of the motes, and therefore extend the lifetime of the network, and to provide scalability to the application. The student will have the opportunity to use the WSN testbed of the Machine Learning Group, which currently makes available one hundred sensor nodes (mainly TMote Sky climatic sensor nodes). One part of the thesis will be on the deployment and maintenance of such a network in the computer science department, or in the Greenhouse of the Solbosch cmapus. A second part will be of the design of learning techniques able to reduce the traffic in the network, by means of in-network, distributed machine learning algorithms.

Required skills: Statistical analysis, machine learning. Required interests: embedded systems, Internet of things, networking.

7. Trading challenge (Gianluca Bontempi, Andrea Dal Pozzolo)

Directa Sim, an Italian online trading broker, is organizing a trading challenge with real money for European master students.

The goal of the project is to use machine learning/statistical techniques to build a trading model. The people interested in participating are supposed to form a group and choose a leader who will manage the trading operation.

For more information: http://www.universiadideltrading.it/index_fr.html or contact Andrea Dal Pozzolo at adalpozz@ulb.ac.beRequired skills: Statistical analysis, machine learning.

8. Sentiment analysis on Facebook (Yann-Aël Le Borgne and Gianluca Bontempi)

Social Media have gained momentum and are used regularly by hundreds of millions of users. Facebook, the third largest web property online, is king of all social networking sites. Visits to Facebook curently account for one in every seven minutes Internet users spent online in October 2011, and 75 percent of the overall time spent on social networking sites.

The subject of this master thesis will be to investigate possible visualization and automatic text mining, as well as sentiment analysis, for the Facebook page of the European Parliament (EP). The page is currently supported by more than 350000 fans and receives on average more than 1000 comments and interactions per day. In its quest for a better understanding of citizen’s concerns, the EP is interested in tools that would allow to visualize and extract trends from this stream of comments. These include for example hot topics extraction, categorization of positive/negative sentiments from the users, visualization through wordclouds or others.

The design of these tools will rely on text mining, machine learning, and sentiment analysis techniques. The main part of the thesis will be to design a system that allows to search and index comments, and to extract trends from the user’s comments. The second part is more exploratory and will imply the design of tools allowing to detect patterns (groups of people with different interests for example) and to vizualize the data (by means of wordclouds for example).

The thesis will be carried out in collaboration with the European Parliament and Digicracy SPRL.

Required skills: machine learning, text mining, sentiment analysis, interest for social networks.