PROJ-H-402 - Computing Projects

This page details the projects proposed by the Machine Learning Group for the computer programming course PROJ-H-402 [WebPage].

1) Navigating a photo album through hand-gesture recognition with the Microsoft Kinect (Yann-Aël Le Borgne and Gianluca Bontempi)

1) Navigating a photo album through hand-gesture recognition with the Microsoft Kinect (Yann-Aël Le Borgne and Gianluca Bontempi)

The Kinect is a sensing device which was launched on the market by Microsoft in November 2010. One of its key features is a depth camera, allowing full 3D motion capture. Since its release, the Kinect has received a tremendous success, holding the Guinness World Record of being the "fastest selling consumer electronics device" after 8 million units were sold in only 2 months.

The Kinect opens the way to a wide range of new applications, in particular in the domain of human-computer interfaces. The 3D camera is accurate enough (resolution of 640*480 at 30Hz) to capture in real-time the 3D of a moving human body, allowing the design of natural software user interfaces based on gesture and posture recognition.

The aim of the project will be to design a prototype of a gesture recognition system allowing a user to navigate through a photo album. 

An idea of what can be achieved can be seen in this video. Given the time constraints associated to the course, the goal of the project is less ambitious than what is advertised in the video, and will consist in designing a program allowing photos to be moved from left-to-right and right-to-left using hand gestures, and where the central picture can be zoomed in and out. Extensions will of course be welcome!

OpenNI, a fast-developing C++ (with Java, Visual Basic, and other languages' wrappers) open source application programming interface which provides high-level functions for gesture recognition, will be used for the implementation.

Requirements: Good or standard knowledge of C++ (other languages possible, Java, VB or Processing), interest for machine learning and signal processing. 

Working languages: English, French.

Contact Persons: Yann-Aël Le Borgne, Gianluca Bontempi

2) 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 networksare numerous, in domains such as ecology, medicine, industry, agriculture, or defense.  

The project 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). The main part of the project will be on the deployment and maintenance of such a network in the computer science department, or in the Greenhouse of the Solbosch campus. A secondary part will be on the design of learning techniques able to reduce the traffic in the network, by means of in-network, distributed machine learning algorithms.

Requirements: Good or standard knowledge of C and Java, interest for machine learning and signal processing. 

Working languages: English, French.

Contact Persons: Yann-Aël Le Borgne, Gianluca Bontempi

3) Natural security : Simulating the co-evolution of MHC proteins and parasites (Tom Lenaerts)

"Major histocompatibility complex (MHC) genes are the most polymorphic genes found in vertebrates, having hundreds of alleles in human populations and, generally, dozens of alleles in other vertebrate species.  Such extreme polymorphism is thought to be associated with the function of MHC proteins, which present antigens derived from parasites to lymphocytes, thus inciting the adaptive immune response. Parasites are under evolutionary pressure to evade detection by the host immune system and are, therefore, expected to adapt most quickly to those MHC alleles that are found at the highest frequency in the host population. The negative frequency-dependent selection resulting from this situation gives a selective advantage to rare MHC alleles and can cause high levels of polymorphism to be maintained in host MHC genes. Another mechanism that may result in maintenance of polymorphism in MHC genes is the heterozygote advantage resulting from an increased range of antigen types that may be presented by MHC heterozygotes compared to homozygotes." (quotation from M.J. Ejsmond, W.Babik and J. Radwan (2010) BMC Evolutionary Biology).