Join us at our next event.

Topic: Applications of Machine Learning in Industry R&D

Next in our spring seminar series, we welcome Dr. Douglas Brooks from Southwest Research Institute (SWRI).

Abstract:
This discussion will highlight past and present projects, showcasing the approaches and results to solving industry focused R&D challenges. The discussion will specifically highlight the unique efforts within the R&D sector. Additionally, the discussion will include some potential future concepts and thoughts.

Those who cannot attend in-person are welcome to tune in virtually through Webex: https://utsa.webex.com/utsa/j.php?MTID=m568fe3d73989cb26d676f896e952c3c0.

Topic: IoTFinder: Efficient Large-Scale Identification of IoT Devices via Passive DNS Traffic Analysis

Next in our spring seminar series, we welcome Roberto Perdisci, Patty and D.R. Grimes Distinguished Professor in Computer Science; and Director, Institute for Cybersecurity and Privacy, School of Computing at the University of Georgia.

Abstract:
Being able to identify and enumerate potentially vulnerable IoT devices across the Internet is important, because it allows for assessing global Internet risks and enables network operators to check the hygiene of their own networks. Towards this goal, in this talk I’ll present IoTFinder, a network-based system for efficient, large-scale passive identification of IoT devices. To develop IoTFinder, we leveraged distributed passive DNS data collection, and developed a machine learning-based system that aims to accurately identify a large variety of IoT devices based solely on their DNS traffic fingerprints. IoTFinder can efficiently detect IoT devices independently of whether they reside behind a NAT or other network middleboxes, or whether they are assigned an IPv4 or IPv6 address. We designed IoTFinder as a multi-label classifier, and evaluated its accuracy in several different settings, including DNS traffic collected at a US-based ISP hosting more than 40 million clients. The experimental results show that our approach allows for accurately and efficiently detecting many diverse IoT devices, even when they are hosted behind a NAT and their traffic is “mixed” with traffic generated by other IoT and non-IoT devices hosted in the same local network.

Those who cannot attend in-person are welcome to tune in virtually through Webex: https://utsa.webex.com/utsa/j.php?MTID=m568fe3d73989cb26d676f896e952c3c0.

Topic: Robotic Neural Networks Predicting the Change in a System

Next in our spring seminar series, we welcome Emily Hannigan from the Robotic Manipulation and Mobility Lab at Columbia University.

Abstract:
Robots have the potential to aid humans from search and rescue work to space exploration, to in-home assistance for those with limited mobility. One of the main milestones required to achieve these ends is a robust ability to manipulate objects and locomote in cluttered and changing environments. To achieve this, a robot needs an internal model of the dynamics of itself and the world: given the current state of the world (the robot and the environment), how will any given sequence of actions change the state of the robot and its environment? This predictive model allows a controller to test out actions “in its head” before actually taking any actions. One of the most promising avenues for creating robust and accurate forward models is to use a neural network to approximate the dynamics of the system. However, the state of the art still falls short in a few areas. While it is easy to learn a data driven dynamics model to predict the change of a system over a short horizon, predicting farther into the future is still an open challenge. Additionally, the more complex a robotic system is, the more variables the model needs to keep track of and predict (we call this the state or observation of the system); the higher the dimensionality of the state, the more data and computation power we need to learn a model. This work investigates ways to improve the long term accuracy of data driven dynamics models using temporal data. It also evaluates methods of reducing state space dimensionality by learning this lower dimensional representation alongside a forward model using a method we dub “Context Training”. Throughout the development of these ideas, we also argue for improved evaluation metrics for learned forward models.

Those who cannot attend in-person are welcome to tune in virtually through Webex: https://utsa.webex.com/utsa/j.php?MTID=m568fe3d73989cb26d676f896e952c3c0.

Topic: Tackling Missing Labels in ML for Security

Next in our spring seminar series, we welcome Dr. Baris Coskun, Principal Scientist at Amazon Wed Services.

Abstract:
Lack of reliable labelled data remains to be one of the major challenges for applying ML to cybersecurity problems. In this talk, Baris will present how recent unsupervised methods can be adopted to alleviate part of the problem. He will discuss how autoencoder architectures can be used to identify suspicious activities in compute environments. He will also talk about how these ML models can be evaluated in the absence of ground truth data using weak signals.

Those who cannot attend in-person are welcome to tune in virtually through Webex: https://utsa.webex.com/utsa/j.php?MTID=mab715a7149604891f64ce657baef84e0.

Topic: Neuromorphic Computing for AI at the Edge

Next in our spring seminar series, we welcome Dr. Cory Merkel, Assistant Professor, Department of Computer Engineering, RIT.

Abstract:
Artificial intelligence (AI) has become the linchpin in a growing number of products, services, and research programs which are aimed at automating and enhancing the human decision-making process.  There is no doubt that AI will play a central role in the future of healthcare, transportation, manufacturing, and defense, among others.  However, the rapidly growing size, weight, and power (SWaP) cost of AI algorithms limits their deployment on devices with practical computing and energy constraints (satellites, wearables, wireless sensors).  In this talk, I will discuss our lab’s research to bridge this gap and enable AI in the most SWaP-constrained environments.  This research takes a holistic approach, examining the entire AI stack, from devices and circuits to algorithms and applications.  At the lowest level, I will present our research on memristor-based circuits for implementing weighted communication pathways in artificial neural networks (ANNs).  Memristors reduce the power and latency associated with running ANNs on traditional computer architectures by directly emulating both the memory and computation of biological synapses.  In addition, memristor plasticity enables on-chip learning and allows ANNs to function in the presence of hardware defects and process variations.  Moving up the design hierarchy, I will discuss research on ANN topologies with partially random connectivity, which can lead to reduced hardware overhead and training cost while achieving state-of-the-art performance on classification tasks.  Finally, the talk will highlight some recent research related to the trustworthiness and potential security vulnerabilities of AI hardware.

Those who cannot attend in-person are welcome to tune in virtually through Webex: https://utsa.webex.com/utsa/j.php?MTID=mab715a7149604891f64ce657baef84e0.

Topic: The deterrence of a “nuclear” September 11^th: Solutions of Artificial Intelligence in enhancing Nuclear Security of urban areas

Next in our spring seminar series, we welcome Dr. Miltos Alamaniotis, Assistant Professor, Department of Electrical and Computer Engineering at the University of Texas at San Antonio.

Abstract:
The terrorist events of September 11 led to a redefinition of the security architecture and its priorities for deterring terrorist acts. Among the scenarios that emerged was the use of nuclear materials to carry out terrorist attacks with wide-ranging consequences. The recent developments in the field of Artificial Intelligence, among many, are also applicable in addressing challenges in the field of security of nuclear energy systems and facilities. The data produced by the supervisory means – e.g. sensors – are multiplying exponentially due to the increasing use of new information and communication technologies, resulting from the sensitive nature of these systems. The use of artificial intelligence techniques offers solutions for the development of high-definition and high-speed analysis of data related to the use, storage and transport of nuclear materials. In this talk, the role of artificial intelligence in upgrading the security architecture against a new 9/11 carried with radioactive and nuclear materials will be discussed. In addition, AI-based solutions and techniques for the early detection and localization of nuclear threats will be presented, while societal concerns arising from the inappropriate use of AI in nuclear security will be analyzed.

Those who cannot attend in-person are welcome to tune in virtually through Webex: https://utsa.webex.com/utsa/j.php?MTID=mbd9ddfab95075e07427696090f3b9bac.