主題： 

Coordinated Autonomous Driving in 5G via Joint Visual 3D Exploration

Associate Chair, Department of Electrical and Computer EngineeringUniversity of Washington, Seattle, WA, USA

摘要：

With the huge amount of networked video cameras available everywhere nowadays, such as the roadside surveillance cameras and the vehicles/drone cameras for autonomous driving or aerial surveillance applications, there is an urgent need of systematic and coordinated mining of the detected video objects in the 3D physical world. Thanks to the ultra-reliable low-latency communication (URLLC) capability of the emergent 5G mobile networks, the information derived from advanced video object detections, multiple object tracking and 3D object localization based on these static and moving cameras can be jointly explored by the mobile edge computing (MEC) and real-time shared by all the local users through V2X infrastructure. To achieve this goal, several critical challenges need to be effectively overcome, more specifically, reliable SLAM-based visual odometry for pose estimation (self-calibration) of moving cameras, robust tracking-by-detection for detected object associations in presence of missing or erroneous detections, reliable ground plane estimation for 2D to 3D inferences, finally coordinated mining of multiple cameras. In this talk, I will cover all these topics and propose our optimized strategies of integrating these research components.

主題： 

摘要：

主題： 

Harnessing Artificial Intelligence

摘要：

With the rise of Artificial intelligence, numerous governments, academic institutions and enterprises have launched various initiatives to capitalize on the opportunities that arise from advances in Artificial Intelligence. In this talk, I will give an overview of the research activities and research goals two academic institutes, the Vector Institute for AI and the Waterloo AI Institute, as well as one industrial institute, the Borealis AI Institute funded by the Royal Bank of Canada.

主題： 

地點：

新竹國網中心 

時間：

2018年12月4日 下午14:00~16:00

講者：

Dr. Rafael Vescovi, Narayanam “Bobby” Kasthuri

University of Chicago, Argonne National Laboratories

摘要：

主題： 

地點：

新竹國網中心 

時間：

2018年9月21日 上午10:00~12:00

講者：

prof.許文翰

高效能與科學計算技術研究中心, 國立臺灣大學

摘要：

主題： 

地點：

新竹國網中心 

時間：

2018年8月7日 上午10:00~12:00

講者：

Mr.柯智仁

中央研究院生物多樣性研究中心

摘要：

主題： 

SDN/NFV-Based Security On Demand System

地點：

新竹國網中心 

時間：

2018年7月31日 上午10:00~12:00

講者：

Dr. 周立德

中央大學資訊工程系 特聘教授

中央大學電算中心 主任

摘要：

主題： 

地點：

中國醫藥大學

台中國網中心

時間：

2018年7月11日 上午10:00~12:00

講者：

prof. Keisuke Nakayama

Kobe University

摘要：

主題： 

Data Applications in the Political Field

數據在政治領域的應用

地點：

新竹國網中心 

時間：

2018年6月13日 上午10:30~12:30

講者：

Dr. 謝一平

思為策略研發長

摘要：

從數據資料的角度，觀察分析人類行為 金融、體育、政治……等。對於資訊爆炸的時代，數據資料研究政治行為 例如選舉投票、議題攻防、網路社群、街頭運動，本演講分享數據如何在政治領域上的應用

主題： 

The shape of Data: A Short Introduction to Topological Data Analysis

資料的形狀：拓樸資料分析之探究

地點：

新竹國網中心 

時間：

2018年5月31日 上午10:00~12:00

講者：

Mr. Frédéric Chazal

Inria Research Center Sacla

摘要：

主題： 

地點：

新竹國網中心 

時間：

2018年5月28日 上午10:00~12:00

講者：

Ms. 許美鈴

嘉義基督教醫院

摘要：

面對老齡化的未來，醫療與養護的整合將成為全球的趨勢。當醫療服務逐漸走出醫院的圍牆，雲端與行動將是支撐服務的必要資訊建設，整合、大數據與人工智慧將為人類的醫療與照護帶來重要突破。但來自不同來源的資料如何整合? 服務團隊彼此間的資料如何互通? 是整合服務與大數據建構必需面對的問題，本演講將與大家分享如何邁向醫療照護的互通世代。

摘要：

In this webinar we will describe the research goals and computational capabilities of Lifemapper software. The Lifemapper platform takes as input georeferenced point data of known species occurrences based on biological museum specimens and observations of species in the wild.  Lifemapper generates predicted species distribution models or suitable habitat models for plants and animals based on correlations between distributions documented by specimens, and environmental data layers such as climate, substrates, and vegetation cover.  In addition the Project computes multi-species distribution models for regions or continents into data structures known as presence-absence matrices which are then used to compute various biodiversity indices and statistics. These derived products are also useful for analyzing the relative contribution of historic biogeography and phylogeny on patterns of species diversity found in any particular area, region, or island.

主題： 

從菜鳥警察到資料戰警 --- Data scientist as police

地點：

新竹國網中心 

時間：

2018年4月13日 上午10:00~12:00

講者：

Mr. 柯維然

新竹市警察局

摘要：

1. 從資料看交通安全
2. 警政資料分析
3. 應用深度學習於車流預測
4. 監錄系統智慧化與邁向人工智慧

主題： 

New Developments in German and European HPC

地點：

新竹國網中心 

時間：

2018年3月28日 上午10:00~12:00

講者：

Dr. Michael M. Resch

HPC Center Stuttgart (HLRS), Germany

摘要：

主題： 

地點：

時間：

2018年3月20日 下午02:00~03:00

講者：

Prof. S. Felix Wu

 Professor of Computer Science and Associate Dean of Academic Personnel and Research, College of Engineering, at UC Davis.

摘要：

主題： 

地點：

時間：

2018年1月25日 下午02:00~05:00

講者：

Dr. Satoshi Matsuoka

Professor, Tokyo Institute of Technology and Fellow, Advanced Institute for Science and Technology (AIST), Japan

Director, Joint AIST-Tokyo Tech. Open Innovation Lab on Real World Big Data Computing

摘要：

With rapid rise and increase of Big Data and AI as a new breed of high-performance workloads on supercomputers, we need to accommodate them at scale, traditional simulation-based HPC and BD/AI will converge. Our TSUBAME3 supercomputer at Tokyo Institute of Technology became online in Aug. 2017, and became the greenest supercomputer in the world on the Green 500 ranking at 14.11 GFlops/W; the other aspect of TSUBAME3, is to embody various Data or "BYTES-oriented" features to allow for HPC to BD/AI convergence at scale, including significant scalable horizontal bandwidth as well as support for deep memory hierarchy and capacity, along with high flops in low precision arithmetic for deep learning. Furthermore, TSUBAM3's technologies will be commoditized to construct one of the world’s largest BD/AI focused and "open-source" cloud infrastructure called ABCI (AI-Based Bridging Cloud Infrastructure), hosted by AIST-AIRC (AI Research Center), the largest public funded AI research center in Japan. The performance of the machine is slated to be several hundred AI-Petaflops for machine learning; the true nature of the machine however, is its BYTES-oriented, optimization acceleration in the memory hiearchy, I/O, the interconnect etc, for high-performance BD/AI. ABCI will be online Spring 2018 and its archiecture, software, as well as the datacenter infrastructure design itself will be made open to drive rapid adoptions and improvements by the community, unlike the concealed cloud infrastructures of today. Finally, transcending from FLOPS-centric mindset to being BYTES-oriented will be one of the key solutions to the upcoming "end-of-Moore's law" in the mind 2020s, upon which FLOPS increase will cease and BYTES-oriented advances will be the new source of performance increases over time in general for any compputing.

主題： 

地點：

時間：

2017年11月22日 上午10:00~11:00

講者：

Mr. Handol Kim

D-wave System Inc.

摘要：

Kanokvate Tungpimolrut was born in Bangkok, Thailand, in 1968. He received the B.Eng. degree in electrical and electronics engineering from King Mongkut’s Institue of Technology Ladkraband, in 1989, and the M. Eng. as well as D. Eng. degree in electrical and electronics engineering from Tokyo Institute of Technology, in 1992 and 1995, respectively.

Following receipt of the D.Eng. degree, he was a researcher with Fuji Elctric Co., Ltd. Since 1996, he has been a researcher with National Electronics and Computer Technology Center, Thailand. His research interests are the induction motor drive system and switched reluctance motor interests are the switched reluctance motor and drive system as well as motor drive applications.

主題： 

地點：

時間：

2017年11月8日 10:am

講者：

Dr.  Rafael Vescovi

Argonne National Laboratory/University of Chicago

摘要：

At low resolution scales (e.g. mm voxels), magnetic resonance imaging (MRI) techniques allow for in vivo or ex vivo mapping of neuronal tracts through a combination of diffusion-weighted imaging techniques and post-imaging computational tractography. While powerful, they cannot achieve the nanometer-scale resolution required to identify neuronal connections. Moreover, they have never been thoroughly validated against ground truth high-resolution data. At the highest resolution scale (e.g. nm voxels), recent efforts at automated electron microcopy (EM) provide synapse-level resolution (3 nm) but are currently limited to brain volumes of the size of a grain of sand and face serious computational challenges scaling to even 2 pixels of MRI data at 1 mm resolution (i.e. almost 2 petabytes of EM data). This unresolved mismatch between these imaging modalities (and others) partitions our understanding of the brain into ‘silos’ divided by resolution scales with little to no cross validation. The strengths of each modality are not leveraged for a more comprehensive understanding of the brain. This problem is only exacerbated since brains potentially operate at multiple scales in parallel (from communication via individual connections between neurons to communication between brain regions). Multi-resolution multi-modal brain maps are critically necessary for a more complete understanding of the brain.
We propose to use synchrotron-based micro-CT (uCT) to fill this gap, bridging the resolution divide between MRI and EM by providing intermediate resolution (e.g. micron voxels) over entire brains and with sample preparation conditions compatible with MRI and automated EM on the same brains. We propose to use uCT as a ‘Rosetta Stone’ enabling mapping of complete neuronal paths, allowing, for the first time, validation of dtMRI, and identifying areas of interest in both coarser resolution modalities for subsequent nanometer reconstructions with automated serial EM.  
 

主題： 

Personalized Transfer Learning

摘要：

In several application domains, data instances are produced by a population of individuals that exhibit a variety of different characteristics. For instance, in activity recognition, different individuals might walk or run with different gait patterns. Similarly, in sleep studies, different individuals might exhibit different patterns for the same sleep stages. In telecommunication networks, software applications might generate packet flows between servers according to different patterns. In such scenarios, it is tempting to treat the population as a homogeneous source of data and to learn a single average model for the entire population. However, this average model will perform poorly in recognition tasks for individuals that differ significantly from the average. Hence, there is a need for transfer learning techniques that take into account the variations between individuals within a population. In this talk, I will describe online algorithms to transfer knowledge on the fly from specific individuals within a population to a new individual in order to bootstrap the learning process in sequential tasks such as activity recognition, sleep stage identification and packet flow prediction in telecommunication networks.

主題： 

環境變遷下的水域生態毒理與生態系代謝: 高頻監測資料與數理模式的應用

地點：

時間：

講者：

摘要：

Dr. Tsai  study demonstrates that terrestrial loads of CDOM serve as a

摘要：

We recently initiated a project to develop a cloud-based machine learning system based on the FCHA. Flow in Cloud is a pool of multiple kinds of processing engines such as FPGA and GPU connected by a circuit switch network. A special operating system, Flow OS, will combine, connect, and provide such engines to users based on job requirements. There are many technology developed behind the scene. One of them is the container technology. This talk will elaborate this effort for future AI platform.

摘要：

股價操縱旨在使價格偏離基本價值並伺機從中獲利，影響資本市場運行、投資人權益與公共利益甚鉅，現存文獻中股價操縱的理論模型眾多，然而相關的預警卻一直未有適當的發展。本文嘗試根據財金理論結合計量方法出一個新的股價操縱預警方式，其利用不同的操縱手段中使股價呈現緩步上漲 (running-up) 的共同現象做為認定條件，精進監理實務中的相關做法。我們以1994 年至 2010 年之間違反證券交易法第 155 條業經法院裁定股票操縱的案例公司為樣本。研究證實本文的操縱預警效果可補足現行的注意股票公告，同時其所認定的股價操縱期間與法院審理所判定的期間也有相似或可供比較的表現。本文的警示方法不僅具有前瞻性、即時性，同時僅需要股價資料而使偵測成本低等優點。後半部討論到如何依據機率模式推估報稅資料中是否出現作假帳的事證。

摘要：

Dr. Shao gives a general background of Biodiversity and its role in Biological informatics. He then deliberates how Taiwan's biodiversity database developed, how the international movement on open data impacts on the biodiversity and how Taiwan contributes and complements to the international biodiversity community. There are a group of datebase and website services, such TaiBNET(TaiCOL)、TaiBIF、TaiBOL、TaiEOL, subsequently developed for online services since 2001. The most recent one is TaiBON starting services from 2015. The more than decade enduring efforts of Dr. Shao on integrating Taiwan's biodiversity data and provide easy and open access to the data help significantly lay the foundation of Taiwan's long term scientific query on biodiversity.

摘要：

Dr. Kasthuri developed an automated approach to large-volume serial electron microscopy ("connectomics"). The Kasthuri lab continues to innovate new approaches to electron microscopic based connectomics reconstructions including making samples more amenable to automatic segmentation and combining proteomic and genomic approaches with electron microscopy. We are also now exploring the use of high-energy X-rays from synchrotron sources for mapping brains in their entirety. The Kasthuri lab is applying these techniques to developing, adult, and aged brains in service of answering the question: How do brains grow up and age?

摘要：

In this webinar we will describe the research goals and computational capabilities of Lifemapper software. The Lifemapper platform takes as input georeferenced point data of known species occurrences based on biological museum specimens and observations of species in the wild.  Lifemapper generates predicted species distribution models or suitable habitat models for plants and animals based on correlations between distributions documented by specimens, and environmental data layers such as climate, substrates, and vegetation cover.  In addition the Project computes multi-species distribution models for regions or continents into data structures known as presence-absence matrices which are then used to compute various biodiversity indices and statistics. These derived products are also useful for analyzing the relative contribution of historic biogeography and phylogeny on patterns of species diversity found in any particular area, region, or island.

摘要：

Dr. Tanaka explains the current organizational morale on pursuing innovative ICT and related human factor in AIST, which includes the research focuses and resources investment. The most prominent one is the AI bridging Cloud Infrastructure (ABCI), the largest investment on AI platform from Japan government, which is claimed to be able to achieve 130~200 PF in AI fashion. The facility will be used to strengthen the competitive edge of AI innovation for Japan.

摘要：

The rise of big data science has created new demands for modern computer systems. While floating performance has driven computer architecture and system design for the past few decades, there is renewed interest in the speed at which data can be ingested and processed. Early exemplars such as Gordon, the NSF funded system at the San Diego Supercomputing Centre, shifted the focus from pure floating point performance to memory and IO rates. At the University of Queensland we have continued this trend with the design of FlashLite, a parallel cluster equiped with large amounts of main memory, Flash disk, and a distributed shared memory system (ScaleMP’s vSMP). This allows applications to place data “close” to the processor, enhancing processing speeds. Further, we have built a geographically distributed multi-tier hierarchical data fabric called MeDiCI, which provides an abstraction very large data stores cross the metropolitan area. MeDiCI leverages industry solutions such as IBM’s Spectrum Scale and SGI’s DMF platforms. Caching underpins both FlashLite and MeDiCI. In this talk I will describe the design decisions and illustrate some early application studies that benefit from the approach.

摘要：

Wind resources in the Taiwan Strait are rich and wind farms have been scheduled to be licensed to wind energy operators. For large scale adoption of windfarms, understandings of the interactions between windturbines and local weather system are hence becoming important. To address the interactions, we apply the weather research forecast (WRF) model to Fuhai, one of the wind farm areas offshore to the Changhua County.  The applicabilities and capabilities of the model will be inspected and the topics will include wind farm wakes, typhoon simulations. The results will be validated by comparison to the wind mast measurements and simplified theoretical models. Along the line, miscellaneous related topics will also be informally addressed.

摘要：

Nature-inspired metaheuristic algorithms, like the particle swarm optimization and many others, enjoy fast convergence towards optimal solution via a series of inter-particle communication. Such methods are common for the optimization problem in engineering, but few in statistics problem. It is especially difficult to implement in some fields of statistics as the search spaces are mostly discrete, while most metaheuristic methods require continuous search domains. This talk introduces a new method called the Swarm Intelligence Based (SIB) method for optimization in experimental design problems within both discrete and continuous spaces. In specifi c, the supersaturated designs (SSD) , Latin hypercube designs (LHD) and minimum energy designs (MED) are optimized. The SSD optimization is served as a demonstration of the standard framework of the SIB method. The LHD optimization shows how the framework is extended to multiple objective problems. The MED optimization shows how the SIB method is used in continuous domain and how efficient if the initial particles are preselected. Ones can modify their own algorithms from the standard SIB method to tackle their own problems, like community detection in social network analysis, change-point analysis in functional analysis, feature selection in classi fication problem, etc.

摘要：

Modeling is an important research field in computer graphics. Based on spatial grammars, we developed a HTML5 visualization interface for modeling of Chinese traditional architectures. Spatial grammars are implemented using Object-Oriented design patterns. With the help of spatial grammars, modeling of wooden structures no longer needs to be done manually. Instead, our system can generate traditional architectures automatically.

請於此頁註冊，此次演講線上登入方式將會寄至您信箱。

摘要：

In this talk, we present a participatory urban sensing project for PM2.5 monitoring. The key feature of this project is its open architecture, which is based on the principles of open hardware, open source software, and open data. By working closely with government authorities, industry partners, and maker communities, we have constructed an effective ecosystem for participatory urban sensing of PM2.5 particles. Based on our deployment achievements to date, we provide a number of data services to improve environmental awareness, trigger on-demand responses, and assist future government policymaking. The project is highly scalable and sustainable with the potential to facilitate the Internet of Things, smart cities and citizen science in the future.

請註冊此頁以登入線上會議室

摘要：

Social Media is changing many different aspects of our lives. By participating in online discussions, people exchange opinions on various topics or contents, shape their stances, and gradually build their own characteristics. In this talk, we will present a framework for identifying online user characteristics and understanding the formation of user deliberation and bias in online newsgroups. Under the SINCERE.se (Social Interactive Networks: Conversation Entropy Ranking Engine), we have designed a dynamic user like graph model to recognize user deliberation and bias automatically in online newsgroups. We evaluated our identication results with linguistic features and implemented this model under SINCERE as a real-time service. By applying this model to large online newsgroups, we study the influence of early discussion context on the formation of user characteristics. Our conclusion is that the formation of user deliberation and bias is a product of situations, not simply dispositions: confronting disagreement in unfamiliar circumstances promotes more consideration of different opinions, while recurring conflict in familiar circumstances evokes close-minded behavior and bias.