Speaker: Senay Solak
John F. Smith Memorial Professor and Department Chair Operations and Information Management Department, Isenberg School of Management, University of Massachusetts Amherst

When: 02:00 AM – 03:00 PM, April 7th, 2023

Abstract:

Information security is an inseparable operational component for any business that utilizes information systems. Given this significance, firms are increasingly concerned about the cost-effectiveness of their investments in information security. In this study, we seek to address two key decisions by a firm related to such investments: how much the firm should invest in information systems security, and how this investment should be allocated over different categories of security countermeasures. As part of our findings, we derive a simple functional relationship between the potential total losses of a firm and the optimal amount that the firm should invest in information systems security. Related to this, we find that firms in the finance, energy, and technology sectors should invest twice as much in trying to detect information security breaches, than in trying to prevent them. In other industries, information security investments should be split evenly between preventive and detective measures. Moreover, the overall information security budgets for certain types of firms in the former set of industries should be on average 15% higher than other industries, even when the potential total losses under a security breach are the same. As some additional conclusions, we find that the value of these optimal policies is higher for small to medium-sized firms, while a gradual investment strategy over a budget period is better than early utilization of the budget at the beginning of this period.         

Flyer for Event: Link

Speaker:
Mark Chen, Research Scientist, Open AI

When: 11:00 AM – 1:00 PM, April 7th, 2023

Abstract:

This talk will provide an overview of large language models (LLMs) and surrounding generative technologies, their development, and applications across various domains. It is tailored for a technical audience with limited exposure to Artificial Intelligence (AI). I will begin with a brief history of language modeling, outlining key developments that have culminated in the emergence of advanced LLMs such as ChatGPT. Subsequently, I will introduce the fundamental principles that drive the efficacy of LLMs, focusing on their ability to lead and represent complex language patterns. I will discuss the technical challenges involved in scaling LLMs as well as the remarkable properties that have emerged at scale, such as the ability to tackle new tasks with no or minimal new training data. I will further introduce representation learning through language modeling and its role across a range of applications. Beneficial AI systems must be both useful and harmless. ChatGPT, a cutting-edge language model developed at OpenAI, was trained using Reinforcement Learning from Human Feedback (RLHF) and other alignment techniques to optimize its behavior and safety. I will give a high-level overview of RLHF and its importance in creating safe and beneficial AI. I will conclude with LLMs applications across other domains, such as DNA and protein sequence modeling.         

Flyer for Event: Link

Speaker:
Prof. Fei Gao

When: 11:00 AM – 1:00 PM, April 7th, 2023

Abstract:

This lecture focuses on Real-time simulation and Hardware-in-the-Loop (HIL) techniques that emerge as indispensable tools in the design, modeling, testing, and validation of modern power electronic systems. Different from an offline simulation, real-time simulation depends not only on the results of logical/arithmetic calculations of the model but also on physical time (real-time) when these results are produced. Today, high accuracy real-time simulation and HIL of modern power electronics is still a great challenge. Power electronics components models are characterized by their high switching frequency and on-off characteristics. Mathematical models that contain a number of power electronic components can lead to serious real-time simulation problems (e.g. limited simulation time step size, large memory to store circuit topology due to on-off states of power switches, etc.). Solve such mathematical models while meeting the strict time constraint requires the development of specific real-time simulation techniques.In this lecture, a general introduction of real-time simulation and Hardware-in-the-Loop (HIL) technology is first given. State-of-art modeling approaches for power electronic switches, as well as electric network formulations of power electronic systems are then presented, and their advantages and drawbacks are discussed with commercial solution examples. The main hardware implementation platforms (CPU and FPGA) for real-time simulation models are also compared. Finally, a simple example about how to implement a Floating Interleaved Boost Converter (FIBC) model in real-time simulation platform for controller Hardware-in-the-Loop (HIL) application is provided using the previously presented techniques.          

Flyer for Event: Link

Zoom Link: https://fiu.zoom.us/j/92361748808?pwd=Q2lHVnRZaUw3UUtYWUhTbGFaSkZRZz09

Panelists:
Dr. Allan Ward (Department of Energy, Solar Energy Technologies Office)

Prof. Patrick McClusky (University of Maryland)

Prof. Sudip Mazumder (University of Illinois Chicago)

Mr. Miles Russel and Mr. Matt Ursino (Yaskawa Solectria Solar)

When: 3:00 PM – 4:00 PM, January 31, 2023

Host: Professor Arif Sarwat (Florida International University)

The theme of the Session:

Reliability is a critical aspect of power electronics and solar PV systems as it ensures that these
systems can function without fail over an extended period. Power electronics are used to convert and control the flow of electrical energy, and their reliability is crucial for maintaining a stable power supply in various applications such as renewable energy systems and electric vehicles. Solar PV systems, on the other hand, rely on the reliability of photovoltaic panels and inverters to convert solar energy into usable electrical energy. The reliability of these components is critical for the overall performance and lifespan of the solar PV system. The system must be able to withstand harsh environmental conditions and maintain a high level of efficiency to provide reliable and sustainable power. Regular maintenance and monitoring of these systems can help ensure their reliability over time.
Inverter failure accounts for a significant portion of solar plant maintenance costs, up to 50%. This is partly due to the time and effort required to fix issues with solar inverters, which leads to power and revenue loss for end customers or project developers. For example, a 2–3-week inverter shutdown in a year can result in a 5% loss of energy and revenue. To improve reliability and mitigate these issues, an integrated approach is needed that includes analyzing field data, using Physics of Failure (PoF) models to assess component degradation and failure, and determining the effects of different operating conditions on component stress. One way to achieve this is to create a test-bed inverter and a digital twin of that inverter, which can then be used to assess reliability after periods of aging and degradation.
Developing a digital twin of a power inverter can be a valuable tool for understanding reliability. The digital twin can use field failure data to identify and classify failure mechanisms and failure site groups based on different environmental and operational conditions. It can also use stress-defining algorithms to convert environmental and operational exposure into electrical, chemical, mechanical, and thermal stresses acting on critical components in the inverter. By combining this information with physical knowledge of the effects of these stresses on key components, the digital twin can then specify key failure mechanisms and components and use physics of failure algorithms to calculate the level of degradation and the time to failure for these components under different conditions. The digital twin can also use these models to determine the level of degraded performance or time to failure for the entire
inverter for different failure mechanisms.

Flyer for Event: Link