In the last decade reinforcement learning has demonstrated tremendous progress in terms of being a model-free control paradigm for decision making in complex systems with uncertainty and partial observability, thus making it a candidate technology for demand response with a promise of true scalability.

We study theoretical problems of fault diagnosis in circuits and switching networks, which are among the most fundamental models for computing Boolean functions.

Ultrawide bandgap (UWBG) semiconductors including AlN, Ga2O3, c-BN, diamond have attracted enormous interests. They offer markedly larger figures of merits for power and RF applications than other known semiconductors.

The talk will introduce the 5G advanced standard vision for vertical industries and relate it to the latest prototyping effort released by the Qualcomm Research teams.

Deep neural networks (DNNs) are revolutionizing computing, necessitating an integrated approach across the computing stack to optimize efficiency. In this talk, I will explore the frontier of DNN optimization, spanning algorithms, software, and hardware. We'll start with hardware-aware neural architecture search, demonstrating how tailoring DNN architectures to specific hardware can drastically enhance performance.

In this talk, I will present an innovative framework we developed to address these limitations and accurately predict pedestrian crossing intentions. At the core of the framework, we implement an image enhancement pipeline to enable the detection and rectification of various defects that may arise during unfavorable weather conditions. Following this, we employ a transformer-based network with a self-attention mechanism to predict the crossing intentions of pedestrians. This pipeline enhances the model's robustness and accuracy in classification tasks. We assessed our framework using the famous JAAD dataset. Performance metrics indicate that our model achieves state-of-the-art results while ensuring significantly low inference times.

In my lecture, I will share personal insights on transitioning from academia to management consulting and creating startups. I will discuss how to assess if a career in management consulting suits your goals and skills, and provide practical advice on interviewing successfully with consulting firms.

The rapid growth of Artificial Intelligence (AI) and Deep Learning mirrors an infectious phenomenon. While AI systems promise diverse applications and benefits, they bear substantial security and privacy risks. Indeed, AI represents a goldmine for the security and privacy research domain.

In this talk, we consider Bayesian parameter inference associated to a class of partially observed stochastic differential equations (SDE) driven by jump processes. Such type of models can be routinely found in applications, of which we focus on the case of neuroscience.

Over the past decades, Jeff has been lucky enough to follow some of these developments from a front-row seat. He will share selected perspectives on current and past research including the various pendulum swings over the years between extremes, such as expertise-based knowledge engineering vs. data-centric machine learning, autonomous vs. mixed-initiative systems, publicly available chatbot capabilities vs. personalized, policy-governed multi-agent systems. In this talk, I will show what past research in AI has to teach us about the future.

Coffee Time: 15:30 - 16:00. The phase-field model, a powerful modeling tool for dealing with interface problems, has been widely used in various fields such as computational physics, computational biology, materials engineering, and even image processing. The dissipation of free energy is an important and fundamental property of the phase-field model.

We provide error estimates and stability analysis of deep learning techniques for certain partial differential equations including the incompressible Navier-Stokes equations. In particular, we obtain explicit error estimates (in suitable norms) for the solution computed by optimizing a loss function in a Deep Neural Network approximation of the solution, with a fixed complexity. This is a joint work with A. Biswas and J. Tian.

Let us consider an elliptic equation of second order in variational form i.e. div(A(x)∇u) = divf in a bounded domain Ω ⊂ Rn where the function f belongs to some suitable function space.

Machine learning (ML) has witnessed remarkable advancements in recent years, demonstrating its effectiveness in a wide array of applications, including intrusion detection systems (IDS). However, when operating in adversarial environments, ML-based systems are susceptible to a range of attacks.

Splitting methods have been shown to a useful tool in solving phase field equations. However, rigorous nonlinear stability analysis has not been available. In this short course, we will discuss some recent development in this direction.

Coffee Time: 15:30 - 16:00. Splitting methods are a well-established tool for numerically integrating time-dependent partial differential equations. These methods split the vector field into disjoint components, which are integrated separately using an appropriate time step. The individual flows are then combined to obtain the desired numerical approximation.

This talk presents a method for obtaining good accuracy of the measured rotational speed and discusses the control algorithms and modulation technique of the high-speed BLDC motor.

Our society keeps entrusting ICT systems with high value cyber-only assets, such as our most sensitive data, finances, etc. However, when it comes to cyber-physical systems and their ability to act in and with the physical world, lifes are at risk and require rigorous protection against accidental faults and cyberattacks.

Light can be controlled through different degrees of freedom. An optical field is described through frequency, amplitude, phase, polarization, and wave-front structure. Many applications have been explored using these degrees of freedom, and some have great importance in our daily life.

Many problems in applied geometry amount to the solution of a typically nonlinear partial differential equation. We will discuss why it may not be a good idea to discretize the equation, but to take the viewpoint of discrete differential geometry and discretize the theory.

In the first part of this thesis, we have discussed the Cram ́er-Rao lower bound (CRLB) to evaluate the performance of beam tracking for a joint beam tracking and symbol detection scheme in deep-space optical communications.

A fascination with symmetric forms seems to be an innate feature of human perception and for millennia it has influenced art and natural philosophy. The concept of symmetry is one of the very few on which mathematicians and physicists agree, namely that Symmetry ≡ Groups. We describe some special symmetries and related problems including symmetric polynomials and monstrous moonshine.

We will give an overview of the development of Abstract Algebra from Galois to our time. The talk will be accessible to general audience with basic mathematical background.

The studies in numerical approximation of partial differential equations are characterized by the necessity of managing complex geometries and their discretization. We focus our attention on two different fields where complex geometries are very common: the mathematical modeling of fluid-structure interaction problems and the family of virtual element methods.

We develop a derivative-free global minimization algorithm that is based on a gradient flow of a relaxed functional. We combine relaxation ideas, Monte Carlo methods, and resampling techniques with advanced error estimates. Compared with well-established algorithms, the proposed algorithm has a high success rate in a broad class of functions, including convex, non-convex, and non-smooth functions, while keeping the number of evaluations of the objective function small.