Data augmentation is a popular technique to considerably improve training of neural networks in various applications, such as computer vision and language modeling.

Abstract

We present a deep adaptive sampling method for solving PDEs where deep neural networks ar

Abstract

We design and implement PG, a Byzantine fault-tolerant and privacy-preserving multi-senso

Abstract

Harmonic functions in Euclidean space are characterized by the

Abstract

With the advancement of supporting technologies, robots are increasingly deployed in a wi

With the advancement of supporting technologies, robots are increasingly deployed in a wide range of applications, from smart factory automation and precision agriculture to large-scale environmental monitoring, such as underwater exploration.

Abstract

Sobolev homeomorphisms play a fundamental role in geometric function theory, calculus of

In this talk, we discuss the weak limits of Sobolev homeomorphisms and their injectivity properties. We show that these mappings are almost everywhere injective when the Sobolev exponent p>n−1.

Abstract

Numerical software is being reinvented to provide opportunities to tune dynamically the a

Abstract

"In recent years, the rapid expansion of model and data scales has notably enhanced the p

Abstract

This talk will give an overview of the research of the High-Performance Visualization res

Autonomous driving technology promises to revolutionize transportation systems, yet faces significant challenges in 3D perception, real-time computing, and decision-making in complex, dynamic environments.

Free boundary problems emerge naturally in mathematical models representing physical, biological, or financial phenomena, such as ice melting, population dynamics, or stock market behavior. These problems involve solving partial differential equations for both an unknown function and an unknown domain. This talk will explore several free boundary problems and different methods to address them.

We introduce a parallel hybrid approach for Bayesian inference of large spatio-temporal Gaussian processes, combining domain decomposition with the Rao-Blackwellized Monte Carlo estimator. This method enhances speed and scalability by integrating iterative Krylov methods with direct factorizations, improving accuracy and robustness in large-scale datasets.

In the realm of fast and scalable approximated Bayesian Inference, two highly sought-after approaches have traditionally been the Laplace Method and Variational Bayes.

With the advent of wearable sensors and internet of things (IoT), there is a new focus on electronics which can be bent so that they can be worn or mounted on non-planar objects.

Second-order partial differential equations (PDEs) are traditionally classified as being parabolic, elliptic, or hyperbolic in nature, and this classification largely determines the kind of analytical and numerical techniques that can be successfully applied to them.

This talk is devoted to the finite element approximation of boundary value problems with interfaces. We shall consider a simple second order elliptic equation with discontinuous coefficients and present the most common approaches for its finite element discretization.

Currently, attention mechanism becomes a standard fixture in most state-of-the-art NLP, Vision and GNN models, not only due to outstanding performance it could gain, but also due to plausible innate explanation for the behaviors of neural architectures it provides, which is notoriously difficult to analyze.

In a society progressively increasingly welcoming automation, robotics constitutes a very active field of research. This presentation will feature recent robotic-related activities taking place at KAUST, ranging from abstract and analytical to concrete and experimental.

The design of efficient parallel/distributed optimization methods and tight analysis of their theoretical properties are important research endeavors. While minimax complexities are known for sequential optimization methods, the theory of parallel optimization methods is surprisingly much less explored, especially in the presence of data, compute and communication heterogeneity.

Deep Neural Networks (DNNs) have demonstrated exceptional performance in various fields, but they often suffer from performance degradation when encountering domain shifts, common in real-world applications. This thesis addresses this challenge by exploring and improving robustness approaches.

The design of efficient parallel/distributed optimization methods and tight analysis of their theoretical properties are important research endeavors. While minimax complexities are known for sequential optimization methods, the theory of parallel optimization methods is surprisingly much less explored, especially in the presence of data, compute and communication heterogeneity.

This seminar examines the transformative impact of wearable sensor technologies across healthcare, environmental monitoring, and marine biology. In healthcare, these sensors enable continuous monitoring for enhanced patient care and disease management.

Large language models (LLM) have become part of our daily life and are now indispensable tools for conducting research as well. The performance of LLMs is known to increase as the model size and data size is scaled up.