Applications of Convolutional Neural Networks for Classification across Multiple Imaging Domains
Dept of Computer Science, Sonoma State University
Machine learning (ML) has become a central technique to solve a wide range of real-world problems. The availability of large data sets combined with the high-speed implementation of algorithms that use various models of the data to reveal the relationship between the input(s) and output(s) makes this approach attractive in all aspects of computational problem-solving. It is used in web search, spam filters, image tagging, and recognition, interactive voice-activated programs like Apple’s Siri, in software that beats human professionals in chess, Jeopardy, Go, etc. Recent deep learning (DL) algorithms have further revolutionized the field and their applications are in diverse fields of medicine, finance, weather, earth sciences, etc. In particular, learning from images has been made possible by constructing deep, layered, and hierarchical models of data, typically represented using Convolutional Neural Networks (CNN). In this talk, I will present research conducted at SSU based on CNNs on projects with varying imaging domains and classification goals:
1) Automatic Classification of Wild Animals from Camera Trap Images
2) Identification of Interstitial Lung Disease from Computed Tomography (CT) scans
3) Classification of Sigma Clasts in Microscopic Photomicrographs of Rocks.
Make Ray Tracing Great Again!
Sacramento State University, Sacramento, CA
In the field of 3D graphics, ray tracing is a technique for generating stunningly realistic lighting effects. However, it also places a heavy demand on computing resources. In other words, it's slow! But in 2018, NVIDIA unveiled hardware ray-tracing support on its Turing microarchitecture consumer GPUs. Now ray-tracing has become a hot topic again, and so this talk will demonstrate programming techniques for achieving ray tracing using compute shaders. The speaker is the author of the "Computer Graphics Programming in OpenGL" series of textbooks for C++ and Java, published by Mercury Learning.
Machine learning and data-driven solutions for cost-efficient network automation
Department of Computer Science, Sonoma State University
Machine Learning (ML) and data-driven solutions have revolutionized many areas of technologies. Communication technology is also increasingly benefiting from such solutions. Automated network resource management powered by ML and data-driven solutions can help to reduce the cost of connectivity, to free up more bandwidths, to foster innovation on the connected services etc., leading to more connected society and businesses. Many time-consuming and complex tasks of network resource management are being automated; thanks to virtualization of network components, advancements in artificial intelligence, and insights learned from data. Dr. Sabidur's research works with Networks Research Labs at UC Davis and AT&T Labs explore important problems in this area of research. This is an exciting new area of research with potential impact on Edge Computing, IoT, Machine Intelligence, Industry 4.0, Smart City, 6G and beyond.
Next-Generation of HPC Batch Scheduling
Lawrence Livermore National Laboratories
Resource management and batch job scheduling is a crucial part of the software stack of operating capable large-scale supercomputers, enabling multiple users to share the available resources fairly and efficiently. In this talk, I will discuss some of the key challenges in the next-generation of HPC scheduling, which include: managing resources such as power, supporting diverse and complex high-throughput workflows, and efficient utilization of heterogeneous components. I will discuss SLURM and Flux, which are both well-known resource management frameworks for HPC. Flux is a hierarchical next-generation resource management framework that is being actively developed at LLNL. I will also present my ongoing research in power-aware scheduling and variation-aware scheduling with Flux.
Sonoma State University
Artificial Phronēsis (AP) claims that phronēsis, or practical wisdom, plays a primary role in high level moral reasoning and further asks the question of whether or not a functional equivalent to phronēsis is something that can be programmed into machines. The theory is agnostic on the eventuality of machines ever achieving this ability but it does claim that achieving AP is necessary for machines to be human equivalent moral agents. AP is not an attempt to fully describe the phronēsis described in classical ethics. AP is not attempting to derive a full account of phronēsis in humans either at the theoretical or neurological level. AP is not a claim that machines can become perfect moral agents. Instead AP is an attempt to describe an intentionally designed computational system that interacts ethically with human and artificial agents even in novel situations that require creative solutions. AP is to be achieved across multiple modalities and most likely in an evolutionary fashion. AP acknowledges that machines may only be able to simulate ethical judgement for sometime and that the danger of creating a seemingly ethical simulacrum is ever present. This means that AP sets a very high bar to judge machine ethical reasoning and behavior against. It is an ultimate goal but real systems will fall far short of this goal for the foreseeable future.
John P. Sullins is a full professor of philosophy at Sonoma State University. He has numerous publications on the ethics of autonomous weapons systems, self-driving cars, personal robotics, affective robotics, malware ethics, and the philosophy and ethics of information technologies as well as the design of autonomous ethical agents.