<html>
<head>
<meta http-equiv="Content-Type" content="text/html; charset=Windows-1252">
<style type="text/css" style="display:none;"> P {margin-top:0;margin-bottom:0;} </style>
</head>
<body dir="ltr">
<div class="elementToProof ContentPasted0" style="font-family: Calibri, Helvetica, sans-serif; font-size: 11pt; color: rgb(0, 0, 0);">
This is an announcement of Takuya Kurihana's Candidacy Exam. </div>
<div class="elementToProof ContentPasted0" style="font-family: Calibri, Helvetica, sans-serif; font-size: 11pt; color: rgb(0, 0, 0);">
<div class="ContentPasted0">===============================================</div>
<div class="ContentPasted0">Candidate: Takuya Kurihana</div>
<div><br class="ContentPasted0">
</div>
<div class="ContentPasted0">Date: Thursday, June 08, 2023</div>
<div><br class="ContentPasted0">
</div>
<div class="ContentPasted0">Time: 10 am CST</div>
<div><br class="ContentPasted0">
</div>
<div class="ContentPasted0">Remote Location: https://uchicago.zoom.us/j/93461688477?pwd=QWxhRnZ4VEN0OGJZTG9kQndQVDJ5UT09 Meeting ID: 934 6168 8477 Passcode: 453282</div>
<div><br class="ContentPasted0">
</div>
<div class="ContentPasted0">Location: JCL 390</div>
<div><br class="ContentPasted0">
</div>
<div class="ContentPasted0">Title: Towards the democratization of large climate dataset via unsupervised deep learning techniques</div>
<div><br class="ContentPasted0">
</div>
<div class="ContentPasted0">Abstract: Artificial Intelligence (AI) for science (AI4Science) develops cutting-edge AI algorithms and High-Performance Computing (HPC) to advance the frontier of science through the discovery of new scientific knowledge. In climate
science, multispectral satellite instruments have provided petabytes of global cloud imagery over the past decades that capture cloud structure, size distributions, and radiative properties at a near-daily cadence. These observations should help in understanding
cloud responses and trends in cloud behavior in climate science, but the complexity and size of this dataset have left it under-utilized. To aid the challenge, I introduce rotationally invariant cloud clustering (RICC) that combines rotationally invariant
autoencoder and hierarchical agglomerative clustering to generate unique new AI-generated cloud classes. Clusters produced from RICC detect meaningful distinctions between cloud textures, using only raw multispectral imagery as an input without reliance on
location, time, derived physical properties, or pre-designated class definitions. To achieve the democratization</div>
<div class="ContentPasted0">of large climate science data using the data-driven cloud classification approach, I create a unique new cloud dataset, the AI-driven cloud classification atlas (AICCA), which clusters 23 years of ocean images from the Moderate Resolution
Imaging Spectroradiometer (MODIS) on NASA’s Aqua and Terra instruments –190 million of roughly 100 km x 100 km patches (128 x 128 pixels) - into 42 AI-generated cloud class labels. AICCA translates 872 TB of satellite images into 56 GB of class labels, metadata,
and 13 cloud physical parameters. AICCA permits the discovery of new classes based on both cloud morphology and physical properties that are unbiased by artificial assumptions and that capture the diversity of global cloud types. Finally, I design a universal
workflow to evaluate the bias exhibited by simulated clouds from high-resolution cloud models, aiming to mitigate the substantial uncertainties associated with climate projections via RICC and AICCA.
</div>
<div><br class="ContentPasted0">
</div>
<div><br class="ContentPasted0">
</div>
<div class="ContentPasted0">Advisors: Ian Foster</div>
<div><br class="ContentPasted0">
</div>
<div class="ContentPasted0">Committee Members: Ian Foster, Rick Stevens, <span class="ContentPasted1" style="font-family:游ゴシック,sans-serif; font-size:14px; orphans:auto; text-align:justify; widows:auto; text-decoration:none; display:inline!important">and Elisabeth
Moyer</span></div>
<br>
</div>
</body>
</html>