In this presentation, I will focus on two primary areas. First, continuing from my technical presentation, I will argue for the study of amorphous solids as a paradigmatic example of a difficult yet structured problem that sits at the forefront of machine learning research on feature extraction, transfer learning, generalization, and interpretability. I will discuss a data-science driven project with the goal of producing highly stable, “aged” glassy systems in silico.
Understanding the behavior of amorphous solids is one of the outstanding challenges in modern condensed matter physics; knowing whether and how to connect local disordered structures with local dynamics is a key difficulty in constructing theoretical descriptions of these materials.
Continue from the MyShake talk on day 1, I will talk about my research vision for the projects I want to do in the near future. Starting from the extension ofMyShake project, a couple of new projects will be introduced in this talk ranging from earthquake early warning, a large-scale sensor array to the bridging of computer science and earth science. All these will be under the umbrella of combining the power of data science and earth science.
In this talk, I will take you on a journey how we have built a global smartphone seismic network - MyShake. This is a crowdsourcing project that harnesses the sensors inside the smartphones to detect earthquakes. For the first part of the talk, I will discuss the motivations behind the project and the steps we took to address the key challenges to build the network. In the 2nd part of the talk, I will talk the data we collected after the release of the application on Feb 12th 2016. We have more than 300,000 downloads and built a global network across 6 continents.
A major societal issue in the Pacific Northwest is the impending large earthquake along the Cascadia subduction zone. Part of my future work is in the development of flexible computational methods for the study of earthquake cycles on geometrically complex faults, specifically for understanding earthquake hazards in Cascadia.
Deficient understanding of the earthquake cycle is the single greatest barrier to minimizing the devastating effects of earthquakes on society and the human environment.
This talk will cover my research in AI, with a focus on Multi-Agent Systems, for solving real-world societal problems, particularly in the areas of Sustainability, Public Safety and Security, Cybersecurity, and Public Health. In these problems, strategic allocation of limited resources in an adversarial environment is an important challenge which involves complex human decision making, a variety of uncertainties, and exponential action spaces.
The dramatic growth in the volume of data and the disproportionately slower advancements in memory scalability and storage performance have plagued datacenter and high-performance computing application performance in the last decade. Emerging heterogeneous memory technologies such as nonvolatile memory (NVM) promise to alleviate both memory capacity and storage problems; however, realizing the true potential of these technologies requires rethinking of software systems in a way that it hasn’t before.
Any given data set can be valuable beyond its original purpose. In this talk, I will provide examples of different data reuses that lead to substantial scientific breakthroughs and discuss how we can increase data reuse in neuroscience. Topics covered will include reducing barriers to data sharing via innovative web repositories such as NeuroVault, providing incentives to share through data papers as well as data analysis as a service (OpenNeuro), and development of standards for neuroimaging data (Brain Imaging Data Structure).
New technologies allow us to understand many biological processes at the molecular level but require principled machine learning methods to capture the underlying dynamics of the cell populations. In this talk, I present two projects. In the first project, we design a dynamic graphical model to jointly analyze different types of genomic aberrations from multi-location/multi-time biopsies of metastatic breast cancer. The model allows us to accurately characterize genomic aberrations and understand oncogenic processes from next-generation sequencing data at a significantly larger scale.