210206 從數據閱讀建築、環境和能源 - 以台灣能源數據為例

Title:
從數據閱讀建築、環境和能源 - 以台灣能源數據為例

Speaker:
傅群 (Chun Fu), PhD student, National University of Singapore

Time:
02/06 (Sat.) 4:30 pm PST, 5:30 pm MST, 6:30 pm CST, 7:30 pm EST
02/07 (Sun.) 8:30 am Taiwan

Keywords:
building, built environment, building science, data science, machine learning, energy consumption

Abstract:
近幾十年來，建築能源預測一直都是研究的熱點。無論是簡單的傳統統計模型到複雜的機器學習模型，或是屬白盒子(white box)的物理模型到屬黑盒子(black box)的純數據驅動(data-driven)模型，都提供了不同角度的能源預測思維。在2019年底，ASHRAE在Kaggle平台上舉辦了”ASHRAE - Great Energy Predictor III”機器學習競賽，能源數據集包含了全球15個城市的近1500棟建築物用電。競賽目標是提出最能準確預測能源的模型方案，本演講會分享此競賽的過程和優勝方法回顧。另外，如今大量收集和開放的數據越來越多，如Google trends和氣象預報資料，本演講也會分享這些數據如何使得建築能源模型更加準確。演講的最後，會以台灣電力公司的能源開放數據搭配氣象局的氣象預報，實作展示一個簡單的機器學習預測服務，並且與官方提供的能源預測服務作比較。

210130 Atom Probe Microscopy and Its Latest Applications

Title:
Atom Probe Microscopy and Its Latest Applications

Speaker:
陳翊昇 Yi-Sheng (Eason) Chen, PhD, The University of Sydney

Time:
01/30 (Sat.) 5 pm PST, 6 pm MST, 7 pm CST, 8 pm EST
01/31 (Sun.) 9 am Taiwan

Keywords:
Materials Science, Microscopy, Hydrogen Embrittlement, Hydrogen Energy, Materials Characterization

  本次演講未提供錄影

Abstract:
三維原子探針(Atom Probe Tomography)是一種近年來快速發展的材料顯微術，這種技術能提供原子級的材料化學成分分布圖，被廣泛運用在各種材料科學研究當中。近年來APT在偵測效率以及先進樣品製備技術上有快速的發展，這些發展促成了APT被應用於了解金屬的氫脆現象，近5年來講者在此領域中數度取得重要突破，於2017年與2020年各有一篇論文獲刊在《Science》期刊上。此演講將會簡介APT技術原理與近年應用發展，期望Tyra聽眾能夠對此技術有更多了解，以利建立未來合作。

210123 從衛星看見人：都市變遷、搬家與心理健康─運用遙測數據與機器學習之跨領域研究 From Pixel to People – Using Satellite Imagery to Monitor Urban Dynamic in Relation to Mental Health

Title:
從衛星看見人：都市變遷、搬家與心理健康─運用遙測數據與機器學習之跨領域研究 From Pixel to People – Using Satellite Imagery to Monitor Urban Dynamic in Relation to Mental Health

Speaker:
陳慈忻 (Tzu-Hsin Karen Chen), PhD, Yale University

Time:
01/23 (Sat.) 6 pm PST, 7 pm MST, 8 pm CST, 9 pm EST
01/24 (Sun.) 10 am Taiwan

Keywords:
geography, environmental science, public health, health geography, urban planning, remote sensing, geographic information system, mental health, urban structure, satellite images, artificial intelligence

  為配合本研究發表時程，本次錄影將延遲上傳

Abstract:
By 2050, another 2.5 billion people are expected to live in urban areas. To fulfill the needs of urban population growth, policymakers are planning new housing projects and urban revitalization. As urban dwellers have often been found at higher risk for depression and anxiety, it would be interesting to consider: Does the urban environment in general create stress, or possibly relieve stress? How is living in a ten-story apartment in the downtown associated with mental health differently from living in a suburban townhouse? These questions have been difficult to answer because data on historical urban structures are often lacking. Fortunately, satellite images provide a big data source to quantify urban structure and its subtle changes. Through these interdisciplinary studies I dived in during my PhD, I will introduce “urban form” and focus on the following questions (1) What is it? (2) How to monitor it? (3) What are the implications of it for mental well-being?

210109 Introduction to Particle Transport Simulation - Deep Penetration Problems

Title:
Introduction to Particle Transport Simulation - Deep Penetration Problems

Speaker:
王孟仁 (Vince Wang), PhD, University of Utah

Time:
01/09 (Sat.) 6 pm PST, 7 pm MST, 8 pm CST, 9 pm EST
01/10 (Sun.) 10 am Taiwan

Keywords:
radiation transport, applied physics, radiation transport, monte carlo method, variance reduction, high performance computation

Abstract:
Particle transport is important for radiation dose calculation for the radiation workers. These simulations are typically called deep penetration problems. With extremely low sampling probability, expensive computational resources is required. To solve the problem, variance reduction technique is required. In this talk, traditional and new variance reduction approaches will be presented. Sample problem from real world will be used to demonstrate the effectiveness of the algorithms.