Evolutionary Computation Applied to Neutrino Detection, OSU, 2016-2021
In August 2016, at the Ohio State University, CCAPP, Columbus, Ohio, Kai was co-organiser and presenter at the first-ever Computing in High-Energy AstroParticle Research (CHEAPR) workshop in which Karoo GP (more below) was featured as a tool for improved understanding of complex data, as Evolutionary Computation applied to Astro-particle Physics. Out of this workshop came the application of genetic algorithms to the evolutionary design of improved antennae for Antarctic neutrino detectors, with APS conference presentations in 2018-20, and publication in 2021. With Dr. Amy Connolly and Dr. Carl Pfendner, The Ohio State University, Dr. Stephanie Ann Wissel, Calpoly, and a host of undergraduate and graduate students at both institutions, Kai provides adjunct guidance and review for this collaborative effort.

Publication: Using Evolutionary Algorithms to Design Antennas with Greater Sensitivity to Ultra High Energy Neutrinos
 

Machine Learning at LIGO:

  • Supernovae Candidate Detection, 2018-2020
    With Dr. Marco Cavaglia, Dr. Sergio Gaudio, Dr. Marek Szczepanczyk, and Travis Hansen. Kai assists with the application of Genetic Programming (Karoo GP) to waveform analysis in single and dual detector cases, in search of supernova. Publication: Currently in review at LIGO.
     
  • Machine Learning Applied to Glitch Detection and Mechanical Couplings, 2017-2018
    With Dr. Marco Cavaglia, University of Mississippi, Kai guided the application of Genetic Programming (Karoo GP) to glitch classification and the follow-on discovery of mechanical couplings at LIGO. Funding provided (Award 1707668) by the National Science Foundation. Publication: Finding the origin of noise transients in LIGO data with machine learning
     
  • TensorFlow Enabled Genetic Programming, a comparative study, 2017
    A performance study of Genetic Programming (Karoo GP) applied to four, varied datasets, on both CPU and GPU computational architectures, using standard Python and TensorFlow computational libraries. Publication: TensorFlow Enabled Genetic Programming

 

Machine Learning Applied to Radio Astronomy Data, SKA-SA, 2014-17
In 2016, Kai Staats completed a Masters of Science in Applied Mathematics, University of Cape Town. Kai was supervised by cosmologist Bruce Bassett, Cosmology Research Group, African Institute for Mathematical Science (AIMS), Muizenberg, South Africa. The research team applied machine learning to automate the process of flagging radio frequency interference (RFI, or man-made radio noise) in radio astronomy data at the Square Kilometer Array (SKA). For his MSc research project, Kai developed Karoo GP, a TensorFlow enabled, Python-based Genetic Programming application. He was the first in the world to apply evolutionary computation to the mitigation of anthropogenic noise in radio astronomy.

In September 2017, Kai returned to the SKA to continue his research under former MSc co-supervisor Dr. Arun Aniyan in the application of Convolutional Neural Networks (CNN) to the flagging of RFI in the Very Large Array, CHILES dataset. In the course of this research effort, Staats and Aniyan developed a multi-dimensional accuracy score for the training of CNNs.

Publications: MSc thesis Genetic programming applied to RFI mitigation in radio astronomy; TensorFlow Enabled Genetic Programming
 

Human Off-World Exploration, 2012-13
In the coming decade science fiction will become science fact as humans leave the confines of the Earth-Moon system and make homes among the asteroids, planets, and some day, the stars. By asking questions about who we are and where we come from, we can more clearly look to the future of human space exploration. The intent is to recognize patterns in human social behavior in the context of isolated, small group dynamics such that we might help guide, or at least anticipate the challenges of the next wave of human exploration and evolution.

Learn more …
 

An International Language for Communication, 2009-2011
Built upon a vocabulary of pictographs, iConji is a communication system for people not of just one ethnic, linguistic, or geo-political foundation, but for all people, everywhere. What makes iConji unique and powerful is its ability to transcend language barriers, to bring people together not only in daily communication via Facebook, Twitter, email, and person-to-person messaging, but in conversation around personal and cultural identity as given foundation in language and art.

With the launch of Over the Sun, LLC, Kai Staats and a new team worked to develop iConji, as an experiment in a new kind of communication.

Learn more …
 

Renewable Energy, 2009-2010
With emerging advances and rapid adoption of photovoltaic technologies, the cost/efficiency ratio is improving such that both residential and commercial installations are enjoying a higher power yield and lower overall cost. While solar and wind electric generation are now viable sources of power, the greatest hurdle remains the discrepancy between time of generation and peak grid consumption. This invokes need for massive, efficient power storage in order to truly harness this clean energy source.

In the fall of 2009, Kai Staats worked with associate AJ Patton to investigate large-scale renewable energy systems as a potential power and revenue source for the Native American people in the American Southwest. On a smaller scale, Kai’s own home was the first in Larimer County, Colorado to be both grid-tied photovoltaic and fully battery backed (“1912-2009: From Fuel-stove to Grid-tied“). What’s more, Kai completely rewired his 1912 home as a reconfigurable patch panel in order to redistribute loads for optimal battery life. Kai has installed two small photovoltaic systems in rural Kenya (2008, 2009), and a 4.2KW system for an isolated ranch in the Rocky Mountains (“To Hold the Sun“).
 

Personal Computing, 2006-2008
As CEO at Terra Soft Solutions, Kai Staats studied the evolution of personal computers and PDAs and how they are received by both domestic and international markets. Kai developed a working prototype (which booted from an SSD 3 years before this was the norm), and patent portfolio around a space then only in its infancy. In 2008 Kai traveled to India, Singapore, and the Philippines to learn about the needs of individual and government users in these growing market segments.

“Personal electronics are rapidly evolving to unify multiple devices into single, highly portable PDAs. Cell phones integrate organizational, communication, and personal entertainment applications while laptops offer solid state storage with longer battery life and lighter weight products at a price similar to high-end cell phones.

“Consumers in both established and developing economies are finding the new, less expensive personal computers not only required for success within a more international, global economy, but also within reach of their disposable income.” — Kai Staats, 2008

More under Sketchbox in the Innovation archives …
 

Linux OS and Supercomputing Development, 1999-2008
For a decade, Kai Staats was co-founder and CEO of Terra Soft Solutions (sold to Fixtars Solutions in 2008), a leading Linux operating system developer. Yellow Dog Linux was developed by Terra Soft for the POWER architecture, including the full line of Apple PowerPC (G3, G4, G5) personal computers and Xserves, select Motorola reference boards, IBM POWER and Cell workstations and servers, the Sony PlayStation3 (PS3), and a number of third party embedded system boards built upon IBM and Motorola POWER architecture CPUs.

yum‘, the defacto RPM package manager for roughly half of all Linux operating systems world-wide had its foundation in an early effort by Terra Soft to beat Red Hat to market with the first graphical installer in 1999, then released as “yup” (yellowdog update).

Terra Soft was a top-tier Apple Value Added Reseller, IBM Business Partner, and provider of integrated, turn-key supercomputing solutions with customers at NASA, Department of Energy, Department of Defense, and Universities around the world. It was estimated that as many as 250,000 home users were running Yellow Dog Linux, at any given time. In those years, Kai and his team conducted extensive research and development of both desktop and high performance computing solutions, including the award winning YDL installer, Y-Bio for bioinformatics, Y-FILM in conjunction with Circle-S Studios, and the highly scalable Y-HPC node construction and management suite.

In addition to his position as CEO, head of sales, and manager of product development, Kai was the designer of the desktop environment for YDL and architect of a 3,000 sq-ft supercomputing facility with redundant, 20 ton cooling, designed to house an initial 400 Sony PS3 servers, up to 2000 rackmount units in total.

Learn more …