The School of Science & Engineering is hosting a seminar series for the fall and spring semester. Seminars are scheduled on Fridays from noon to 1:00 pm in room 205. Below is a list of seminar talk announcements with the associated times and dates for the presentation.
This talk will discuss how the features of cloud computing can be combined with machine-generated big data to provide companies, government organizations, and academia with a vast array of cyber security insights to protect your valuable assets, including features not often exploited. Using a multitude of sensor points and advanced analytic engines, cyber defense can be strengthened and insider threat awareness can be expanded-all while improving daily infrastructure knowledge and understanding.
With a dramatic increase in cyber threads over the last decade, government and industry alike have recognized the pressing need to combat cyber attacks on networks and systems threatening disruption of critical operations and exposure of private information. Educational institutions play an important role in 1) researching technology that improve resiliency of network and systems and 2) growing a workforce that understands cyber security challenges and can study and combat cyber attacks. The UWF Cyber Security Battle Lab is a designated space in the School of Science & Engineering designed for collaborative, inter-disciplinary learning and research. The lab is facilitated with state-of-the art network and computing equipment to conduct a range of different experiments on systems and networks in a controlled and isolated environment. It allows faculty to simulate different real-world system configurations and expose students to various experimental settings for hands-on experience in cyber security. This talk describes the lab's infrastructure and the new software tools and tutorials that have been built to study and monitor cyber attacks along with appropriate defense strategies. Results from a number of experiments conducted in the new lab will be presented. The talk concludes with future directions for the Battle Lab space in the broader context of the UWF Cyber Security initiative.
Information discovery has been transformed by the web, and the impact on how information is gathered, published, and delivered has been profound. Web search is one form of discovery, and is preferred when a user has a specific objective. It is less effective when the user intent is weak, typically the case when visiting a website in the browse mode. Examples include news recommendation, product recommendation, movie recommendation, and many more. Algorithmically matching items to users to maximize user satisfaction is essential to scale such large scale systems. Furthermore, these systems are dynamic with both user preferences and item inventory changing over time, hence the matching algorithms must adapt automatically. But feedback to directly measure and optimize customer satisfaction on an ongoing basis is hard to obtain. A practical approach that is often pursued is to optimize readily available proxies like click-through rates, time spent post-click, purchase rates, and so on. Such optimization can be performed by using large scale machine learning and statistical models.
In this talk, I will describe the art of constructing proxies that are typically used by large consumer centric web companies, and the the science of optimizing such proxies through machine learning and statistical methods. The talk would illustrate concepts through real-world applications I have deployed at both Yahoo! and LinkedIn. Other than machine learning and statistics, I will also discuss some of the engineering and business challenges.
"Knowledge Representation" is traditionally seen within the Computer Science community as a means of expressing knowledge that is suitable for machine processing. In this talk we view knowledge representation as a mediating representation between humans, without an aim for formality. We will show through real-world examples how concept maps as a knowledge representation tool can be used by people of all ages and cultures from pre-schoolers to scientists, and how it is a powerful learning and knowledge sharing tool for all domains of knowledge.
Many human-machine tasks could benefit from bidirectional interaction between the system and the operator. The application space drives the process of solution design. For a given environment, the process may include cognitive task analysis, concept mapping and system modeling to identify interaction bottlenecks between the human machine team before ever attempting to implement a solution. This provides an opportunity to investigate the solution space before committing assets to hardware. This guides the creation of a physical embodiment to address the systemic bottlenecks. The final step requires validation of the solution through demonstration of changes in cognitive effort. We will discuss this multistep process in the context of recent projects.
Electric motors look like black boxes with wires and a shaft. They obey to physics laws, particularly Maxwell's laws. Mainly two factors make their design evolve: the new material (permanent magnets, insulation) and the capability to use efficient models. More and more often, motors are designed for a specific application, including the integration in the global system. On the other hand, a motor is in fact an electric drive, including transmission, power and control electronic. If a fine design requests experience and specific knowledge, efficient tools exist in order to make the best pre-choices in a systemic approach. First, it is important to understand that a motor is a torque generator; this torque, in relation with the motor type, defines the motor size. Second, Scaling Laws are a simple methodology which allows understanding better the properties and the limits of specific motors. They also offer the possibility to compare the motor types and the structures, such as internal and external rotor.
Application to electric vehicles implies specific constraints such as mass, volume, reliability, thermal aspects, etc. Generally, a classical vehicle (gas) has 40 to 80 electric motors and activators: weepers, starter, battery generator, air conditioning, fan, door closing, etc. For an electric car one or two power electric motors are added and offer more flexibility and control possibilities. Design criteria are presented according to the applications.