Projects
Unsupervised and semi-supervised learning using kernel methods. KU Leuven
Multi-labeled semi-supervised learning in big data problem Ghent University
In the era of big data, analyzing and extracting knowledge from large-scale data sets is becoming a very challenging task. The application of standard data mining tools in such data sets is not straightforward. Hence, a new class of scalable mining methods that embraces the huge storage and processing capacities of cloud platforms is required.
In this research proposal, we will design highly scalable methods for preprocessing and data ...
Semi-Supervised and Explainable Machine Learning with an Application to the Low-Voltage Grid KU Leuven
Due to the enormous increase in available data and the computational power needed to process it, interest in the use of machine learning (ML) keeps increasing as it is able to extract valuable insights from this data. In an effort to make it easier to use ML in practice, this thesis proposes several novel ML techniques driven by two industry needs: (1) to make machine learning systems understand what the user wants and (2) to make the user ...
The re-use of field reference data in space and time to the mapping of vegetation: the potential of semi-supervised 'and' active learning ' (RE-LEARN). University of Antwerp
Semisupervised category learning KU Leuven
In the human category learning literature, category learning is typically investigated in a supervised or an unsupervised way. Supervised category learning involves that participants receive feedback after each encounter of a category member, whereas unsupervised learning implies that no information about the category label is ever provided. However, both forms of category learning seem ecological implausible. In the current dissertation, we ...
Energy-efficient communications through self-supervised learning and distributed wireless networks KU Leuven
Given the ever-growing demand for high throughput wireless communications, the power consumption used by these networks continues to soar. Many research efforts have been focused on providing higher throughput, reliability and lower latencies. However, energy consumption remains an afterthought rather than a first principal goal in the development of wireless technologies. Given the growing concerns regarding carbon emissions and ...
New Methods for Self-Supervised Image Representation Learning KU Leuven
Representation learning plays a key role in most machine learning algorithms. When solving a particular task, the input is usually mapped to an intermediary latent space before being mapped to the final output space. We refer to the representation of an input as the set of feature values to which the input is mapped in the latent space. A good representation captures the essential information present in the data in a readily accessible ...
Overcoming Complexity of Visual Perception by Efficient Pose Estimation and Self-Supervised Representation Learning KU Leuven
Machine vision is a tremendously complex task considering the extent of semantics an image can convey and the structure of the content in an image. It is infeasible to instruct a machine to understand visual sensory data by defining a set of rules and heuristics. However, thanks to deep neural networks, data-centric approaches have led to many breakthroughs that make machine vision possible in a wide range of real-world environments. ...
Semi-Supervised Models with Deep Architectures and Applications KU Leuven
In many real-life applications, ranging from data mining to machine perception, labeled data are very hard and expensive to attain as it requires expert human efforts. Therefore in many cases one often encounters a large number of unlabeled data whereas the labeled data are rare. Semi-Supervised Learning (SSL) is a framework in machine learning that aims at learning from both labeled and unlabeled data points and thus reducing the human ...