Optimizing Low-Dimensional Structure: Theory and Methods in Manifold Learning
High-dimensional data is ubiquitous in modern information systems, yet most of its
intrinsic structure lives in far fewer dimensions. This talk explores the theoretical
foundations of manifold learning and the optimization principles that underpin
dimensionality reduction techniques. We begin by formalizing the manifold hypothesis
and its implications for data representation, then survey key embedding methods
through a unified optimization lens. Special attention is given to probabilistic and
neighborhood-preserving approaches, examining how their objective functions balance
competing goals such as local fidelity, global structure, and computational tractability.
We conclude with practical insights on method selection, common pitfalls, and open
challenges in learning faithful low-dimensional representations from high-dimensional
data.
In the Era of Generative AI and Machine Learning: Challenges and Opportunities
This talk begins with an overview and brief historical perspective on existing AI and ML models, before turning to the emerging field of generative models. Particular attention is given to the challenges and opportunities arising from integrating representation learning with deep neural networks. While the effectiveness of deep learning in supervised settings is now well established, recent advances demonstrate that these models are equally capable of learning rich and complex representations in both supervised and unsupervised settings. By unifying feature extraction with the learning objective, these approaches enable the development of representations inherently aligned with the task at hand. The seminar concludes with a selection of recent research contributions that highlight a range of real-world applications, including predictive maintenance for aircraft engines, natural language processing for low-resource languages, and applications in the medical field.
Energy Efficiency in Resource-Constrained Networks
This talk addresses energy efficiency in resource-constrained networks. After a brief introduction to the specific limitations of such systems, it provides an overview of the main energy optimization mechanisms. It also explores emerging approaches, including energy harvesting and wireless recharging. Finally, optimization strategies aimed at extending network lifetime are presented, along with perspectives for future research.