Charles L. and Ann Lee Brown Distinguished Seminar Series
We are proud to welcome leaders in electrical and computer engineering and related disciplines as participants in the Spring 2023 Charles L. and Ann Lee Brown Distinguished Seminar Series.
Our speakers are eager to share their insight and expertise relevant to our research strengths, including image processing with a focus on health and medicine; machine learning, signal processing and communications; hardware for artificial intelligence and the internet of things; devices and circuits; robotics and control systems; and the design and integration of materials in novel electronic, thermal and optical devices.
We will conduct the Distinguished Seminar Series in a hybrid learning environment, combining in-person engagement with a Zoom webinar on Friday afternoons from Date to Date.
This talk will give an overview of recent developments in the evolution of the Domain Name System (DNS), the distributed global database that provides name to address mappings (and more) for the Internet. After a brief review of the DNS, it will cover how the worldwide DNS ecosystem has been evolving in recent years, and where it might be going in the future.
Medical AI and Pervasive Sensing: The future of Medicine
While monitoring critically ill patients through manual assessments and diagnoses is still considered the gold standard, manual methods impose severe time and personnel resource limitations. Many critical care indices are currently repetitively assessed by overburdened Intensive Care Unit (ICU) nurses, e.g., physical function.
Quantitative Visualization of Microwave Circuits Under Realistic Operating Conditions
Microwave circuits, especially power transistors, are essential components of mobile communication as they amplify signals to be transmitted wirelessly from the base-station to subscriber terminals. Their compact design is increasingly difficult, as mobile network operators require operation at higher output power and frequency while simultaneously demanding a reduction in the circuit size. Shrinking the device and increasing operational frequency results in significant internal electromagnetic coupling and increased power results in higher temperatures, both of which are detrimental to performance.
Changing Modeling and Simulations Needs for Grid Modernization
Power system analysis has long relied on large scale computer simulations for operation and planning of the grid. These simulations rely on a set of well-understood physical models and generally agreed upon simplifications, such as, the separation of transmission and distribution, ZIP models for aggregate loads, and so on. Recent trends in the grid requires a rethinking of the models and these underlying assumptions.
Nanocomposite materials for energy conversion and storage
Composite materials are made up of two or more constituent materials. Through proper design, composite materials can combine the properties of their components and interfaces to optimize the overall performance. Especially, nanostructured composite materials are interesting due to their short diffusion paths and profound interfaces.
Data and AI Model Markets: From Data Sharing, Discovery, and Integration to Values and Ecosystems
Data and AI model sharing is a long time bottleneck for AI and data economies. In this talk, I will argue that data and AI model discovery and integration are foundations for sharing. I will also revisit why sharing remains a big challenge and why many existing approaches like data warehouses, data lakes, federated databases, and federated learning are still far from enough to solve the problem.
Spectrum Sharing for 5G and beyond: a Network Economics Approach
The evolution of commercial wireless networks to 5G and beyond will continue to increase the demands for wireless spectrum. Traditionally, commercial wireless service providers have utilized spectrum that is exclusively licensed to them.
Offline Reinforcement Learning: Towards Optimal Sample Complexity and Distributional Robustness
Offline or batch reinforcement learning seeks to learn a near-optimal policy using history data without active exploration of the environment. To counter the insufficient coverage and sample scarcity of many offline datasets, the principle of pessimism has been recently introduced to mitigate high bias of the estimated values.
Mind the Gap: Advancing Electronics and Photonics through Materials Research
This presentation will overview National Science Foundation programs aimed at advancing electronics and photonics through materials research. Continued advances in the range and capabilities of computing, communications, and energy technologies, to name a few, presents tremendous challenges and opportunities.
Ferroelectric Devices, Circuits and Architectures for AI Hardware Design
In this presentation, I will present the recent progresses on doped HfO2 based ferroelectric devices such as FeFET. First, I will discuss the fundamental device physics including the minor loop switching and history effect, the drain-erase scheme and the variability and scalability of FeFET.
VLSI Architectures for Training Deep Neural Networks and for Homomorphic Encryption
Machine learning and data analytics continue to expand the fourth industrial revolution and affect many aspects of our lives. The talk will explore hardware accelerator architectures for deep neural networks (DNNs). I will present a brief review of history of neural networks. I will then talk about reducing latency and memory access in VLSI accelerator architectures for training DNNs by gradient interleaving using systolic arrays.
Enabling Self-Powered Monitoring: Advances in Micro-Power Thermoelectric Generators for IoT Sensors and Wearables
The advent of self-powered wearable sensor nodes and electronic devices has engendered a plethora of possibilities for the seamless integration of health and wellness monitoring into the daily lives of individuals. Progress in the realms of energy efficiency and harvesting, along with the concurrent optimization of form factors and augmentation of functionalities, have been particularly salient. A self-powered wearable system typically comprises sensors, an energy harvesting device, a power management unit, energy storage capability, a data transmission component, and a data processing platform.
Learning to read xray: applications to heart failure monitoring
We propose and demonstrate a novel approach to training image classification models based on large collections of images with limited labels. We take advantage of availability of radiology reports to construct joint multimodal embedding that serves as a basis for classification. We demonstrate the advantages of this approach in application to assessment of pulmonary edema severity in congestive heart failure that motivated the development of the method.