2018 研究主題清單 (2018 Research List)


主持人
(PI' Name)
研究主題
(Research Name)
研究介紹
(Introduction)
其他資訊
(Other Information)
王釧茹
Wang, Chuan-Ju
資料表示法學習與其應用

Data Embedding for AI-enabled Applications
學習不同類型資料(如文本,音頻信號,圖片,對象,行為,社交關係和用戶行為)的表示法是 AI 相關應用中的一項重要工作,其可用於許多不同的領域如:電子商務,金融,醫療,製造等。 在這個研究題目中,學生將會學習到幾種表示學習演算法,並運用於有趣的實際應用。

Learning meaningful representations of data (such as texts, audio signals, pictures, objects, actions, social relations, and user behaviors) is a fundamental step for AI-enabled applications across various fields, including E-commerce, finance, healthcare, manufacturing, and agriculture. We need such representations (or embeddings) to gain insights into data, to make predictions, to optimize cost and profit, and to facilitate decision making. In the research topic, students are expected to learn several state-of-the-art representation learning algorithms, and adapt them to interesting real-world applications.
PI個人首頁(PI's Information) :
https://cfda.citi.sinica.edu.tw/~cjwang

實驗室網址(Research Information) :
https://cfda.citi.sinica.edu.tw/


Email :
cjwang@citi.sinica.edu.tw
陳伶志
Chen, Ling-Jyh
以空氣盒子系統為基礎的空品物聯網資料分析研究

Research of AirBox System for Advanced Data Analysis and Innovative Applications
我們透過空氣盒子計畫,已在全球31個國家,布建超過4000台PM2.5微型感測器,其中有超過2000台的機器遍布台灣全島,提供每五分鐘一筆的微型感測數據,已成為台灣目前最重要的民間空氣品質量測資料提供單位,更是全球數一數二的 PM2.5 微型感測與環境物聯網資料中心與系統典範。

在這個暑期專案中,我們希望透過兼具時間與空間高解析度的 PM2.5 感測資料,運用資料科學、人工智慧或其他技術,進行兼具學理、創意與應用價值的資料混搭與進階分析。內容可以是(但並不局限於)即時污染源的溯源、中尺度的 PM2.5 擴散模式推估、中尺度的 PM2.5 濃度預報模式建構、PM2.5 衍生的社經資源成本推估、PM2.5 濃度與即時生理訊號的整合分析,甚或是其他更具創新與挑戰的研究議題。

我們歡迎對本項研究主題有興趣、有想法,並且願意接受挑戰的優秀人才加入我們的團隊,一同學習、努力、並對當前重大的環境議題做出貢獻。

AirBox 計畫網址: https://pm25.lass-net.org/


AirBox has been the largest IoT system for PM2.5 monitoring in the world. Till now, it has installed more than 4,000 units in 31 countries, and the number is still fast increasing. Among them, more than 2,000 devices are installed in Taiwan, and each of them provides micro-measurement of PM2.5 concentrations at a rate of every 5 minutes. AirBox system has become the most popular system and data source of air quality for people in Taiwan, and it has become one of the most successful model of IoT systems in the world.

In this summer project, we wish to utilize the fine-grained and spatio-temporal data of our system, and conduct advanced data analysis with both research and practical values. The project may involve advanced techniques of statistics, data science, and artificial intelligence; and the potential topics include (but are not limited to) PM2.5 emission source tracking, fine-grained PM2.5 dispersion modeling, fine-grained PM2.5 concentration forecasting, social economic impacts of PM2.5 pollution estimation, and the correlation between PM2.5 concentration and physiological signals investigation. We also welcome innovative and even more challenging topics on the related problems.

We are looking for self-motivated, creative, and open minded people to join us. We will learn together, work together, enjoy the process together, and produce good results at the end together. For further questions, please feel free to contact us.

AirBox project: https://pm25.lass-net.org/
PI個人首頁(PI's Information) :
https://sites.google.com/site/cclljj/

實驗室網址(Research Information) :
https://sites.google.com/site/cclljj/NRL


Email :
cclljj@iis.sinica.edu.tw
蘇克毅
Su, Keh-Yih
應用深度學習進行智慧型問答與跨文件處理

DNN-based Intelligent QA and Multi-Document Processing
自然語言理解(Natural Language Understanding, NLU)為人工智慧當前最重要的研究領域之一。其中「機器閱讀」是指電腦能夠自行透過閱讀學習知識(read to learn)、並能以學習的知識來增強自己的閱讀能力(learn to read),本研究室目前致力於建立機器閱讀系統,並應用於智慧型問答以及多文件處理。此研究專題的重點將放在 (1) 把領域背景知識融入深度學習網路(DNN),以增進 DNN 之學習效率;(2) 建立基於知識庫(Knowledge Base)的問題回答系統,加強問題辨識能力與推論能力;(3)結合遠監督式機器學習與 DNN,以期能減少領域知識的訓練資料量,並有效擴增知識庫。

Natural Language Understanding (NLU) is one of the most popular fields among AI studies. One of the major tasks, Machine Reading, enables computers to obtain knowledge from given texts in aids of logic inferences, and aims not only "read to learn" but also "learn to read". We are establishing a machine reading system with applications to intelligent QA systems and multi-document processing. In this project, we expect to integrate domain knowledge into DNN to increase its performance, and apply distant supervision to reduce training data of domain knowledge while improve inference power and enlarge knowledge base as well.
PI個人首頁(PI's Information) :
http://www.iis.sinica.edu.tw/pages/kysu/

實驗室網址(Research Information) :
http://nlul.iis.sinica.edu.tw/research_ch.html
https://

Email :
kysu@iis.sinica.edu.tw
王新民
Wang, Hsin-Min
語音、語言與音樂處理

Speech, Language and Music Processing
我的研究興趣是語音處理、自然語言處理、多媒體資訊檢索、機器學習與模型識別,研究目標是開發多媒體音訊(主要是語音與音樂)分析、抽取、辨識、索引及檢索技術。進行中的研究工作包括自動語音辨識、語者辨識、語音轉換(例如說話人語音轉換、中性語音轉表達性語音)、語音文件檢索/摘要、自動影片配樂、音樂資訊檢索等。

My research interests include speech processing, natural language processing, multimedia information retrieval, machine learning, and pattern recognition. The research goal is to develop methods for analyzing, extracting, recognizing, indexing, and retrieving information from audio data, with special emphasis on speech and music. The ongoing research includes automatic speech recognition, speaker recognition, voice conversion (e.g., speaker voice conversion and neutral speech to expressive speech conversion), spoken document retrieval and summarization, automatic generation of music video, music information retrieval, etc.
PI個人首頁(PI's Information) :
http://www.iis.sinica.edu.tw/pages/whm/

實驗室網址(Research Information) :
http://slam.iis.sinica.edu.tw/


Email :
whm@iis.sinica.edu.tw
劉庭祿
Tyng-Luh Liu
源於GAN的深度學習技術與網路精簡化在電腦視覺的應用

GAN-Inspired Deep Learning Techniques and Network Compression for Computer Vision Applications
本研究規劃旨在發展深度學習技術與其於人工智慧領域的最新應用。有關技術方面,我們將深入研究源於GAN模型的深度學習原理。並建構實用的訓練機制與理論,來有效解決以往學習GAN深層網路的不確定性。為了驗證本計畫所發展出方法的效能,我們將其應用於數個頗具挑戰性的人工智慧與電腦視覺問題,其中包含了視覺與語言、視覺與智慧環境、及視覺與360全景視訊等應用。此外,本計畫的另一項研究重點是發展新的學習機制,可將一個給定的深層網路精簡化。在保有該深層網路原有效能的前提下 將網路參數大幅減少。

In this research project, we aim to develop deep learning techniques for a wide spectrum of emergent artificial intelligence (AI) applications. At the core of our research motivation is to generalize the idea of Generative Adversarial Networks (GANs) so that we could establish new deep learning models and
frameworks to yield effective and practical solutions for challenging computer vision problems. To demonstrate the effectiveness of the GAN related deep learning methods developed in this project, we address several active applications, including 1) Vision and Language, 2) Vision and Smart Environment, and 3) Vision and 360 Video. The other focal point of this project is to establish state-of-the-art algorithms for compressing deep neural networks (DNNs). This aspect of consideration is not only practical in reducing the size of a learned network but also useful in speeding up the run time.
PI個人首頁(PI's Information) :
http://www.iis.sinica.edu.tw/pages/liutyng/

實驗室網址(Research Information) :
http://www.iis.sinica.edu.tw/~liutyng/


Email :
liutyng@iis.sinica.edu.tw
修丕承
Hsiu, Pi-Cheng
嵌入式深度學習於電腦視覺應用

Embedded Deep Learning for Computer Vision Applications
中研院資創中心電腦視覺實驗室 & 嵌入式暨行動計算實驗室 和台大電機/台大資工的研究團隊 (包含Computer Vision、Deep Learning、Embedded Systems、Hardware Chips的團隊) 合作一個跨領域的計畫做Embedded Deep Learning for Computer Vision and Graphics,目標是讓計算能力有限的裝置也有能力做深度學習,執行特定的電腦視覺與圖學應用。學生將由修丕承博士與林彥宇博士共同指導,從事嵌入式系統與深度學習研究。

Deep learning technologies are at the core of the current AI revolution. The affordable GPU hardware and large annotated datasets jointly allow the training of deep learning models with hundreds of layers and millions of parameters. The deep learning-enabled breakthroughs result in great successes in numerous AI research fields. However, deep learning models typically have severe demands on the computing resources of devices, which makes leveraging the power of deep learning on resource-constrained platforms or in real-time applications infeasible. The mission of this project is to address this issue and increase the applicability of deep learning. Students involved will co-advised by Dr. Pi-Cheng Hsiu and Dr. Yen-Yu Lin and gain excellent experience in embedded systems and deep learning research.
PI個人首頁(PI's Information) :
https://www.citi.sinica.edu.tw/pages/pchsiu/

實驗室網址(Research Information) :
http://emclab.citi.sinica.edu.tw/
http://cvlab.citi.sinica.edu.tw/

Email :
pchsiu@citi.sinica.edu.tw
楊奕軒
Yang, Yi-Hsuan
創音計畫:基於深度學習之自動音樂生成系統

GenMusic Project: Music Generation by Deep Learning
參與本實驗室有關音樂生成的研究。對此題目有興趣的人需對於以下至少兩種深度學習演算法有一定的認識和經驗:CNN、RNN、GAN、與VAE。

This project is related to our ongoing music generation project.  Promising candidates should have some hands-on experiences on at least two of the following deep learning algorithms: CNN, RNN, GAN and VAE.
PI個人首頁(PI's Information) :
https://www.citi.sinica.edu.tw/pages/yang/

實驗室網址(Research Information) :
http://mac.citi.sinica.edu.tw/


Email :
yang@citi.sinica.edu.tw
林彥宇
Lin, Yen-Yu
基於深度學習的電腦視覺技術研究

Developing Deep Learning Technologies for Computer Vision Applications
電腦視覺研究旨在讓電腦能像人類視覺系統一般,可感知、識別、甚至解釋它所看到的影像畫面,它是人工智慧領域相當重要的一個分支。深度學習技術,尤其是卷積神經網絡(CNNs),近年在電腦視覺應用中取得了迅速的發展與長足的進步,但目前仍存在一些挑戰和局限性。我與我實驗室助理主要(但不限於)從四個方向發展深度學習技術,以更妥適地解電腦視覺應用,包括:A)融入傳統電腦視覺方法至深度學習模型;B)多模態深層感知;C)開發深度學習技術於無監督式影像共同區域偵測;D)高執行效率的嵌入式深度學習。第一個方向旨在讓深度學習模型透過融合傳統電腦視覺方法而提升效能;第二個方向將利用額外的多模態訊號來輔助與增強電腦視覺應用的效果;最後兩個方向分別試圖緩解深度學習在大量訓練資料以及高計算資源的依賴,以降低深度學習的使用門檻,提高其適用範圍。實習生可依興趣選擇上述一至兩個研究方向或其他相關研究議題進行。

Computer vision research aims to enable computers to see, recognize, and interpret the world like human visual systems. As one of the most important branches of AI, computer vision research has gained very significant progress based on recent advances in deep learning, especially convolutional neural networks (CNNs). Although deep learning technologies are rapidly developed for vision applications in the past few years, there are still some cruxes and limitations. My research assistants and I identify four directions for developing deep learning technologies for computer vision applications, including A) integrating classical computer vision methods into deep learning, B) deep multimodal perception, C) unsupervised common region discovery using deep learning, and D) embedded deep learning for inference acceleration. The first direction aims to make conventional and deep learning-based vision approaches complement and improve each other. The second direction is to leverage extra multimodal signals to facilitate vision applications. The last two directions try to alleviate two main limitations of deep learning, i.e., the requirements of large-scale annotated training data and GPUs, respectively. You might join one or two aforementioned projects, or work on a highly related research topic.
PI個人首頁(PI's Information) :
https://www.citi.sinica.edu.tw/pages/yylin/

實驗室網址(Research Information) :
http://cvlab.citi.sinica.edu.tw/


Email :
yylin@citi.sinica.edu.tw
曹昱
Tsao, Yu
結合聽覺視覺資訊之輔具開發

Assistive Hearing and Speaking Devices based on Integrated Audio and Visual Cues
語音辨識為多感官共同合作的過程,聽覺管道和視覺管道訊息的整合為辨識語音中不可或缺的歷程。聽者藉由「讀話」(speechreading),一邊聆聽語音,一邊觀察說話者的嘴型、臉部表情、手勢等,以獲得完整的語音訊息。而當欲聆聽的語音訊息不完整時,例如當聽者身處吵雜的情境中時,聽覺和視覺的訊息整合能力更有其重要性。研究發現,溝通障礙的族群整合視聽覺訊息的能力呈現困難,而聽力損失或老化亦是造成視聽覺整合能力不佳的影響因素。另一方面,在機器自動語音辨識的研究中亦發現,語音及影像事件的整合能更有效增強語音、提高自動語音辨識的結果。因此,本研究將應用多感官訊息整合的技術,整合語音、嘴型、臉部表情與手勢動作的資訊,提出一套創新型的聽覺輔具,以協助聽損者提升語音辨識的成效。研究成果亦將發展為一套整合多感官訊息之創新型聽能訓練平台,作為聽語專業人員為聽損者進行訓練及評量成效的工具。

我們期望發展整合嘴型、臉部表情與手勢動作以有效提升語音辨識之技術,並且開發聽能評估應用軟體:提供測試者(聽力師或語言治療師)為聽損者進行評量,依評量結果調整助聽輔具設定及聽能訓練平台內容,此應用軟體除了以注音輸出做為評量依據,亦計畫整合腦電波儀(EEG)結果於評測系統,期更直接獲得測試者的辨識結果,避免因測試者輸出錯誤造成評測失真的問題。

Speech recognition is a process that occurs not only in the auditory modality but also the visual modality. The role of visual information is particularly prominent when the auditory speech is contained in a less favorable signal-to-noise ratio. In human communications, the integration of auditory and visual information is a mandatory process and speech disorders may result from difficulties integrating and processing heard and seen speech signals.

Our research is to include visual information to provide enhanced recognition of speech signals for the individuals with hearing impairment. Towards this end, we plan to work on a multimodal framework expanding from the integrated deep and ensemble learning algorithm. The role of visual information in recognizing tonal speech, especially Mandarin Chinese, will be first explored in the human-computer interfaces for devising the optimal parameters to integrate audio and visual information in the proposed framework. We will then evaluate the derived optimization criteria and test the proposed framework structure using objective speech quality index methods and subjective human speech perception tests. It has been verified that brain responses can be used as an indicator to speech perception results. To achieve an unbiased and direct evaluation, we plan to incorporate the Electroencephalography (EEG) measurements into the evaluation system to check the rehabilitation progress and optimize the rehabilitation program. The optimized framework structure will be developed into a multimodal training tool for hearing-impaired individuals to consistently and regularly practice in the speech-language intervention and aural rehabilitation process.

The outcomes of the proposed project will be development of three tools for enhancement of hearing assistive framework, performance evaluation and a multimodal training program for hearing-impaired individuals.
PI個人首頁(PI's Information) :
https://www.citi.sinica.edu.tw/pages/yu.tsao/

實驗室網址(Research Information) :
https://www.citi.sinica.edu.tw/pages/yu.tsao/


Email :
yu.tsao@citi.sinica.edu.tw
黃文良
Hwang, Wen-Liang
適用於影像分析的深度學習演算法

Deep Learning method on Image Analysis
我們將研究探討適用於處理影像分割問題的深度學習演算法

We will study the deep learning method together specifically on image segmentation problems.
PI個人首頁(PI's Information) :
http://www.iis.sinica.edu.tw/pages/whwang/

實驗室網址(Research Information) :
http://www.iis.sinica.edu.tw/pages/whwang/


Email :
whwang@iis.sinica.edu.tw
徐讚昇
Hsu, Tsan-sheng
高效率資料密集計算演算法之設計和實做

Design and Implementation of efficient data intensive algorithms
設計和實做高效率資料密集計算演算法

To solve data intensive computing problems abstracted from applications
PI個人首頁(PI's Information) :
http://www.iis.sinica.edu.tw/~tshsu/

實驗室網址(Research Information) :
http://www.iis.sinica.edu.tw/~tshsu/


Email :
tshsu@iis.sinica.edu.tw
呂俊賢
Lu, Chun-Shien
深度感知

Deep Sensing
壓縮感知(Compressive Sensing/Sampling, abbreviated as CS)是一種新型態的取樣理論。其主要是針對稀疏訊號(sparse signal)可以突破傳統取樣頻率至少必需達到Nyquist rate的限制,CS僅需擷取少量的samples或measurements,在待還原訊號滿足sparsity的條件下,即可利用最佳化的方法來重建原始訊號。CS的特色就是在取樣的同時兼具的降維/壓縮的效果,因此也適用於各種電力或運算能力有限的(行動)裝置。
然而,壓縮感知有多方面的因素(如感測矩陣、dictionary等之設計與訊號具備何種程度之稀疏性或low-rank)需要考量,再加上耗時的最佳化訊號重建;如何賦予壓縮感知更彈性有效的設計與應用則是我們的研究動機。
著眼於深度學習帶來革命性的影響、較少受到人為因素干擾的學習效果、與其快速的inference(計算)能力,我們研究``深度感知(Deep Sensing)'';能以深度學習來實現壓縮感知。
1. 首先,我們研究``深度感知學習網路’’,提出可實現壓縮感知的深度學習網路架構,達成低耗能的資料擷取與低耗能的資料重建;這在過去壓縮感知領域裡是辦不到的。
2. 接著,我們研究``最佳化/壓縮深度感知學習網路’’,主旨在降低整體計算複雜度與網路的大小,並滿足效能的有限損失。這部分的難處在於處理這三個互為牽制的目標仍是不容易。經此最佳化設計後,整個功能便能移植於行動裝置做更廣泛的應用。
3. 最後,對於前兩項研究出的深度感知學習網路,其考量的訊號稀疏性是無結構化的,在壓縮感知文獻中通常假設對欲重建的原始訊號有一定的prior knowledge,如sparsity (重要位置個數)或connected-components個數是已知的。很明顯地,這樣的假設過於理想化,故我們考量結構化(圖形)資料進行``圖形化深度感知學習網路’’之研究。我們的目的有二:(1)利用深度學習幫助壓縮感知自動習得結構化的support pattern而非hand-crafted support set。(2)利用壓縮感知賦予Graph Convolutional Network具備deep sensing能力。如此一來,不僅能大幅降低計算複雜度還能具備scalability (適應於各種網路規模)。

Compressive Sensing/Sampling (abbreviated as CS) is a new paradigm of simultaneous sampling and compression. Without being restricted to the constraint of Nyquist rate, compressive sensing can, in theory, perfectly reconstruct the original signal under the constraints that if only a few samples or measurements extracted from an original signal are available and the signal is sparse. The unique characteristic of CS is that sampling and compression can be simultaneously achieved such that CS is suitably used for resource-limited mobile devices and sensors.
Nevertheless, there are several aspects needed to be considered in CS, including the design of sensing matrix and dictionary/basis, the sparsity or low-rank of signals, and time-consuming optimization for signal recovery. Thus, our motivation is to study the efficient and flexible design of compressive sensing.
With an eye on the revolutional impact of deep learning, we propose studying ``Deep Sensing’’ to achieve deep learning-based compressive sensing.
1. First, we study ``Deep Sensing Learning Neural Network’’ to present a deep learning architecture to realize compressive sensing in that low energy-consuming data acquisition and signal reconstruction can be achieved simultaneously.
2. Next, we study ``Optimized/Compressed Deep Sensing Learning Neural Network’’. The goal is to reduce the computational complexity and network size but at the same time only allow limited performance loss. The difficulty is that it is still not easy to satisfy the compromise among the three requirements.
3. Finally, we will consider structured support pattern of a signal. In the compressive sensing literature, it is usually to assume a prior knowledge (sparsity or number of connected components are known) about the signal to be reconstructed. Obviously, such an assumption is too ideal. Thus, we consider structured (graphic) data and propose ``Graph Deep Sensing Learning Neural Network’’. Our goal is two-fold: (1) Deep learning is used to help compressive sensing to learn structured support pattern instead of hand-crafted support set. (2) Compressive sensing is used to give Graph Convolutional Network(GCN)the capability of deep sensing.
PI個人首頁(PI's Information) :
http://www.iis.sinica.edu.tw/pages/lcs/

實驗室網址(Research Information) :
http://www.iis.sinica.edu.tw/pages/lcs/


Email :
lcs@iis.sinica.edu.tw
宋定懿
Sung, Ting-Yi
癌症蛋白基因體之生物資訊研究

Bioinformatics for Proteogenomics analysis of Cancers
蛋白質是基因最後的產物,在細胞內執行各種不同的生物功能。在醫藥研究方面,蛋白質是最主要的藥物標的。因此在後基因體時代,蛋白體研究也因質譜儀實驗技術的精進而蓬勃發展,癌症研究逐步跨入癌症相關的蛋白體探索。質譜儀為蛋白體研究上最常用的實驗方法,質譜儀產生大量質譜資料,因此質譜資料計算與分析是重要課題。本實驗室是台灣極少數進行蛋白體學生物資訊研究的實驗室,我們專攻蛋白體學研究上質譜儀資料處理之計算方法及軟體系統開發。目前正在執行台灣癌症登月計畫(Taiwan Cancer Moonshot Project),是中研院團隊進行生物資訊研究與資料分析的一員,我們實驗室主要參與肺癌研究。

目前癌症研究以經由蛋白體分析,跨入蛋白基因體分析;以肺癌為例,台灣肺癌病患常可發現epidermal growth factor receptor (EGFR ) 在基因體層次有突變發生,目前需要在蛋白質層次上看到這些突變。我們研究的目標,是希望找出與肺癌相關特定蛋白質中的變異胜肽,並建立一套合適的分析流程,以提高鑑定出的肺癌相關之變異蛋白質的質譜圖的可性度。我們竭誠歡迎有志學習、有熱情的同學加入暑期實習。

Proteins are the final product of genes that execute various biological functions in cells. Furthermore, in biomedicine, proteins are the most prominent drug targets. Therefore, after the genomics era as the advancement of mass spectrometry technology, proteomics research became prevailing and essential in cancer research. Mass spectrometry (MS) is the most commonly used experiment technology to conduct proteomics research. As the advancement of MS technology, high-throughput MS data will be generated. The analysis of such big MS data is an very important topic. Our lab is one of the very few labs conducting research on bioinformatics for proteomics. We particularly work on mass spectrometry data analysis, including algorithm design and software development. Now we are jointly conducting Taiwan Cancer Moonshot Project and working on MS data analysis for lung cancer research.

Nowadays, cancer research further moves from proteomics analysis to proteogenomics analysis. For example, mutations on epidermal growth factor receptor have been observed at genomics level in Taiwan lung cancer patients; and we are interested in finding these mutations at proteomics level. Our goal is to find variant peptides in some specific proteins that are relevant to lung cancer and to develop a data analysis pipeline to facilitate the discovery of variant peptides and their confidence.
PI個人首頁(PI's Information) :
http://www.iis.sinica.edu.tw/pages/tsung/

實驗室網址(Research Information) :
http://ms.iis.sinica.edu.tw/Comics/


Email :
tsung@iis.sinica.edu.tw
王大為
Wang, Da-Wei
資訊科技於醫療與防疫之應用

Information technology for mediecine and epidemiology
應用資訊科技於醫學,公衛與防疫.應用的技術包括: 演算法,機器學習/資料分析, 醫學資訊標準,模擬系統. 應用包括從醫學資料中發現知識,模擬系統用於防疫

Applying information technology to medicine and/or epidemiology.Technology include: algorithm, machine learning/data analytics , medical standards, agent-based simulation. Application include: knowledge discovery from medical data, simulation for disease control.
PI個人首頁(PI's Information) :
http://www.iis.sinica.edu.tw/pages/wdw/

實驗室網址(Research Information) :
http://chess.iis.sinica.edu.tw/lab/


Email :
carol@iis.sinica.edu.tw
陳孟彰
Chen, Meng Chang
PM2.5預測系統

PM2.5 prediction
本研究使用深度學習、大數據分析等技術來進行PM2.5的小間距、長達72小時的預測,以及汙染源偵測。暑期實習生可以選擇深度學習或PM2.5預測來進行暑期實習。

In this research, we apply deep learning and big data analysis to predict fine-qrained, 72-hours PM2.5 prediction, as well as annomaly detection. Summer interns can choose between deep learning and PM2.5 prediction as target of summer project.
PI個人首頁(PI's Information) :
http://www.iis.sinica.edu.tw/pages/mcc/

實驗室網址(Research Information) :
http://ants.iis.sinica.edu.tw/


Email :
mcc@iis.sincia.edu.tw
楊得年
Yang, De-Nian
社群網路與巨量資料分析與新穎應用、多媒體網路最佳化與效能分析

Social Networks and Big Data Analytics and Innovative Applications, Multimedia Network Optimization and Performance Analysis
http://www.iis.sinica.edu.tw/pages/dnyang/descriptions_en.html
http://www.iis.sinica.edu.tw/pages/dnyang/publications_en.html

http://www.iis.sinica.edu.tw/pages/dnyang/descriptions_en.html
http://www.iis.sinica.edu.tw/pages/dnyang/publications_en.html
PI個人首頁(PI's Information) :
http://www.iis.sinica.edu.tw/pages/dnyang/

實驗室網址(Research Information) :
http://www.iis.sinica.edu.tw/pages/dnyang/


Email :
dnyang@iis.sinica.edu.tw
許聞廉
Hsu, Wen-Lian
探索統計式準則於自然語言處理之各項應用

Applications of the statistical principle-based approach on natural language processing
近年來自然語言處理的研究方法主要分為規則式以及機器學習統計方法兩種。然而,規則式方法需仰賴大量人力建構及維護規則,機器學習演算法則需大量的語料作為訓練基礎,這些問題導致目前自然語言處理及理解的研究面臨到不少的困難。本年度我們計畫將針對機器學習演算法的限制所開發的統計準則式方法(Statistical Principle-based Approach, SPBA)應用於當前的自然語言處理及理解研究,包括文獻資料擷取(Reference Metadata Extraction, RME)、中文及生物醫學領域之專有名詞辨識(Named Entity Recognition, NER)及關聯擷取(Relation Extraction, RE)、中文問答系統(Question Answering, QA)、機器閱讀(Machine reading)、小學數學問題解析(Analysis of mathematical problems in primary school)、中文輸入法(Chinese input method)、文件主題偵測(Text topic detection)、意見分析(Opinion mining)等。

Recent approaches for natural language processing can be roughly categorized into two types, namely, rule-based methods and statistical machine learning (ML) methods. The former type relies on a large number of hand-crafted rules, and the latter is heavily dependent on the size of the training corpus. The lack of resources hinders the research on natural language processing and understanding. This year, we focus on the application of the statistical principle-based approach (SPBA), which was developed specifically to avoid the common pitfalls of ML methods. Possible applications include reference metadata extraction, named entity recognition, question answering, machine reading, analysis of mathematical problems in primary school, Chinese input method, text topic detection, opinion mining, etc.
PI個人首頁(PI's Information) :
http://www.iis.sinica.edu.tw/pages/hsu/

實驗室網址(Research Information) :
http://iasl.iis.sinica.edu.tw/


Email :
hsu@iis.sinica.edu.tw
林仲彥
Lin, Chung-Yen
基因體大數據資料解析

Analyze Biomedical Big data for Genome Biology
我們的團隊主要研究模式與非模式生物之多維基因體學(OMICS),包括基因體、轉錄體、單細胞轉錄體與交互作用等巨量資訊數據分析,同時也定序、重組與註解了多個重要經濟生物之基因體,研究團隊並專注於跨領域的研究工作,歡迎不同領域(資訊、統計、數學及生物相關)的人才一起合作。研究範圍以人體幹細胞、水生經濟動物及環境微生物為主,同時發展新的高速計算工具及雲端分析平台,以及引入深度學習等策略,來探討基因、病原與環境的三角互動關係。

The main goal of our team is to analyze omic big data which may lead us to know more about the secrets of biological regulations hidden among massive data deluge.  By combination of open source tools and self-developed programs/ platforms, we have assembled, annotated and decoded the several genome with high economic importance. New approaches like deep learning will be introduced and polished our studies.
PI個人首頁(PI's Information) :
http://www.iis.sinica.edu.tw/pages/cylin/

實驗室網址(Research Information) :
http://eln.iis.sinica.edu.tw


Email :
cylin@iis.sinica.edu.tw
王柏堯
Wang, Bow-Yaw
區塊鏈於隱私透明化上之應用

Applications of Blockchains in Transparent Privacy Protection
本研究將探討區塊鏈應用於隱私透明化保障機制上,相關之理論及實作問題。

We will investigate applications of Blockchains in transparent privacy protection mechanisms.
PI個人首頁(PI's Information) :
http://www.iis.sinica.edu.tw/~bywang

實驗室網址(Research Information) :
http://www.iis.sinica.edu.tw/~bywang


Email :
bywang@iis.sinica.edu.tw
陳郁方
Chen, Yu-Fang
電腦程式自動分析檢測工具的相關研究

Researches on Automatic Program Analysis Tools
我們實驗室的研究目標是讓寫程式變得更容易和更有效率。我們十分歡迎對於寫程式很有熱誠的夥伴加入。目前暑期實習我們有下面的研究目標可以選擇。

1. 大數據處理程式的自動分析與優化(效能與正確性等)
2. 網路程式安全性自動檢測的核心技術
3. 其他相關的研究題目



1. Automatic Analysis and Optimization of "Big-Data" Manipulating Programs
2. Core techniques for the verification of security properties of web-programs
3. Other related research topics
PI個人首頁(PI's Information) :
http://www.iis.sinica.edu.tw/~yfc/

實驗室網址(Research Information) :
http://www.iis.sinica.edu.tw/~yfc/
https://

Email :
yfc@iis.sinica.edu.tw
張佑榕
Chang, Ronald Y.
實現於智慧型手機平台的多感官、多功能的聽覺輔助科技 (智慧聽)

Multi-Modal, Multi-Functional Hearing Assistive Technology on Smartphone Platform (SmartHear)
請見「研究主題英文介紹」

Introduction:
[1] https://www.youtube.com/watch?v=e9HqIj09QJs
[2] https://www.youtube.com/watch?v=gbhFDyH4SVg
[3] Y.-C. Lin, Y.-H. Lai, H.-W. Chang, Y. Tsao, Y. Chang, and R. Y. Chang, "SmartHear: A Smartphone-Based Remote Microphone Hearing Assistive System Using Wireless Technologies," IEEE Systems Journal, in press (DOI: 10.1109/JSYST.2015.2490104).
[4] A. Chern, Y.-H. Lai, Y. Chang, Y. Tsao, R. Y. Chang, and H.-W. Chang, "A Smartphone-Based Multi-Functional Hearing Assistive System to Facilitate Speech Recognition in the Classroom," IEEE Access, volume 5, pages 10339-10351, June 2017.

The candidate will participate in an interdisciplinary project (SmartHear) in one or more of the following aspects: 1. mobile app development/enhancement, 2. implementation of speech-audio processing techniques and multi-functions on Android mobile app, 3. promotion of our system locally and globally for people with hearing loss. Proficiency in Android app programming is required. Knowledge of wireless communications and/or speech signal processing is a plus.
PI個人首頁(PI's Information) :
https://www.citi.sinica.edu.tw/~rchang/

實驗室網址(Research Information) :
https://www.citi.sinica.edu.tw/~rchang/


Email :
rchang@citi.sinica.edu.tw
蘇黎
Su, Li
音樂資訊檢索、人工智慧、音樂互動式系統、訊號處理、計算音樂學

Music information retrieval, artificial intelligence, music interactive systems, signal processing, computational musicology
音樂與文化科技實驗室(Music and Culture Technology Lab)成立於2017年。我們應用最先端的數位訊號處理、深度學習技術在各種結合音樂與人工智慧的問題上,包括自動採譜、聲源分離、機器鑑賞、即時音樂互動系統、音樂自動生成、計算音樂學等等,目標為研發促進音樂文化融入生活的科技。

The Music and Culture Technology Lab was founded in 2017. We apply the emerging digital signal processing and deep learning techniques on the problems combining music and AI, such as automatic music transcription, source separation, machine connoisseurship, real-time music interactive system, music generation, and computational musicology. Our goal is to develop innovative technologies making music culture as a part of our everyday life.
PI個人首頁(PI's Information) :
http://www.iis.sinica.edu.tw/pages/lisu/

實驗室網址(Research Information) :
https://sites.google.com/view/mctl/


Email :
lisu@iis.sinica.edu.tw
古倫維
Ku, Lun-Wei
(1) 多模式聊天機器人對話深度模型 (2) 社群媒體情感與意見分析 (3) 和電腦一起學語言: 自然語言教學工具

(1) Multimodal Social Bot Dialog Generation (2) Sentiment Analysis and Opinion Mining on Social Media (3) NLP Tools for Learning Chinese/English
在這些研究主題中,將學習到自然語言處理之資訊擷取、文章分類、文字生成等概念,另涵蓋自然語言基礎工具的使用及機器學習、深度學習的模型建立等先進技術。可與老師討論希望選擇的研究主題。

Interns will learn how to use basic natural language processing tools, extract information from texts, classify documents, and generate dialogs. Machine learning and deep learning technologies for NLP will be touched. Interns can select the topic/team they wish to join.
PI個人首頁(PI's Information) :
http://www.iis.sinica.edu.tw/pages/lwku/

實驗室網址(Research Information) :
http://academiasinicanlplab.github.io/


Email :
lwku@iis.sinica.edu.tw
鐘楷閔
Chung, Kai-Min
密碼學、複雜度理論或量子密碼學之獨立研究

Independent Research on Cryptography, Complexity Theory, or Quantum Cryptography
The intern is expected to perform independent research on selected topics in (Quantum) Cryptography, Complexity Theory, or general theoretical computer science (TCS) that interest him/her. This often starts by surveying research papers and presenting it to the PI. Along the way, the intern can identify research questions with the PI, perform independent study on the questions, and discuss with the PI in research meetings. Candidate topics include, but not limited to, Quantum Key Distributions (QKD), Post-quantum Cryptography, Lattice-based Cryptography, Differential Privacy, Non-malleable Codes, Device-independent Cryptography,  PRAM Cryptography, Zero Knowledge, Randomness Extractors, etc.

The intern also has opportunity to join our group meeting, and is encouraged to interact with other group members to learn different research topics.

The intern is expected to perform independent research on selected topics in (Quantum) Cryptography, Complexity Theory, or general theoretical computer science (TCS) that interest him/her. This often starts by surveying research papers and presenting it to the PI. Along the way, the intern can identify research questions with the PI, perform independent study on the questions, and discuss with the PI in research meetings. Candidate topics include, but not limited to, Quantum Key Distributions (QKD), Post-quantum Cryptography, Lattice-based Cryptography, Differential Privacy, Non-malleable Codes, Device-independent Cryptography,  PRAM Cryptography, Zero Knowledge, Randomness Extractors, etc.

The intern also has opportunity to join our group meeting, and is encouraged to interact with other group members to learn different research topics.
PI個人首頁(PI's Information) :
http://www.iis.sinica.edu.tw/pages/kmchung/

實驗室網址(Research Information) :
http://www.iis.sinica.edu.tw/~kmchung/


Email :
kmchung@iis.sinica.edu.tw
陳昇瑋
Sheng-Wei Chen
金融科技中的人工智慧應用

AI Applications in FinTech
http://www.iis.sinica.edu.tw/~swc/talk/ai_in_fintech.html

http://www.iis.sinica.edu.tw/~swc/talk/ai_in_fintech.html
PI個人首頁(PI's Information) :
http://www.iis.sinica.edu.tw/~swc

實驗室網址(Research Information) :
http://dirl.iis.sinica.edu.tw/index.html


Email :
swc@iis.sinica.edu.tw
穆信成
Mu, Shin-Cheng
函數編程與程式推理相關問題

Functional Programming and Program Reasoning
我的研究興趣是程式語言與函數編程。在函數語言中,一個程式(理想上)僅是一個數學式,而執行程式就是將其化簡成一個值。這種簡單的計算模型給我們的好處:是我們有了許多好性質,可用來做推理,檢驗程式的正確性,甚至可由轉體規格與需求開始,經由數學方法一步步推導出程式。

本領域可做的大方向包括
* 設計幫助推理用的符號、程式語言、型別系統等。
* 挑選一些演算法問題,嘗試以數學方法實際證明演算法之正確性,或將演算法推導出來。

細節可再討論。

My research interest concerns programming language and functional programming. The fact that a program in a functional language is (ideally) just a mathematical expression, and to run a program is to evaluate it to a normal form. This simple model allows programs to have many nice mathematical properties, with which we may reasoning about programs, prove their correctness, or even derive a program stepwise from its specification.

The general directions of our research could be:

* design symbols, languages, or type systems that aids the programmers in reasoning about programs, and

* pick an algorithm, and apply our approaches to prove its correctness or even to derive an algorithm.

More details can be discussed.
PI個人首頁(PI's Information) :
http://www.iis.sinica.edu.tw/pages/scm/

實驗室網址(Research Information) :
http://www.iis.sinica.edu.tw/page/research/ProgrammingLanguagesandFormalMethods.html?lang=zh


Email :
scm@iis.sinica.edu.tw
黃彥男
HUANG, YEN-NUN
協助 WWW2020 國際學術研討會準備工作

Helping WWW2020 organization and planning
協助準備會議流程,籌備文件, 當地住宿行程及宴會之規劃

Helping WWW2020 meeting agenda, documents,  room and board,  and banquet setup
PI個人首頁(PI's Information) :
https://www.citi.sinica.edu.tw/pages/yennunhuang/

實驗室網址(Research Information) :
https://www.citi.sinica.edu.tw/pages/yennunhuang/index_zh.html


Email :
starmoon6619@gmail.com
蔡懷寬
Tsai, Huai-Kuang
生物資訊

Bioinformatics
我們實驗室的研究興趣是找尋基因體上的蛛絲馬跡,嘗試了解生命體如何透過基因調控系統,成就多采多姿的生命萬象。以資訊技術與統計方法分析生物實驗的大數據資料 (big data),目前研究方向為利用資料探勘與機器學習技術了解基因表現的調控,以及分析生物實驗的相關資料,例如影像或生理反應指標等等。

Our lab's research interests are looking for clues in the genome and trying to understand how the gene regulatory system works in the living organisms. We apply information technology and statistical methods to analyze the big data of biological experiments. The current research directions are transcriptional regulation of gene expression and applications of biological-image analysis by using data mining and machine learning techniques.
PI個人首頁(PI's Information) :
http://www.iis.sinica.edu.tw/pages/hktsai/

實驗室網址(Research Information) :
http://bits.iis.sinica.edu.tw/?id=1


Email :
hktsai@iis.sinica.edu.tw; jiahsin.huang@gmail.com;
何建明
Ho, Jan-Ming
金融科技實務: 以新穎人工智慧演算法解決風險控管  交易決策與投資難題

Practice of Financial Technology: Solving risk management, trading decision, and puzzle of investment with modern artificial intelligence technique
大數據技術的興起使金融技術(FinTech)成為近十年來的熱門話題。
FinTech的難題是市場的波動和不確定性。
在最近十年之前,金融衝擊和危機預測幾乎是不可能的。
幸運的是,機器學習和深度學習的人工智能革命是克服市場預測障礙的機會。
這個項目,我們將通過這些人工智能算法來研究解決信息評級,交易時機,資產定價,利率決定等金融技術問題。

The rising of big data technology make Financial Technology (FinTech) as a hot topic in recent decade.
The puzzles of FinTech are fluctuation and uncertainty of market.
Financial shock and crisis forecasting are almost impossible before recent decade.
Fortunately, the artificial intelligence revolutions which are rose with machine learning and deep learning  are a opportunity to overcome the barrier of forecasting of market.
This project, we will study about solving FinTech issues, such like credit rating, trading timing, asset pricing, interest rate determination, and etc, by these artificial intelligence algorithm.
PI個人首頁(PI's Information) :
http://www.iis.sinica.edu.tw/pages/hoho/

實驗室網址(Research Information) :
http://cscl.iis.sinica.edu.tw/CSCL/default.asp


Email :
jackbaska@iis.sinica.edu.tw
李德財
Lee, Der- Tsai
計算幾何上叢集問題探討

Some Clustering problems on Computational Geometry
考慮d維空間R^d中, 一組n個點址的集合S和另外一組m個點的集合X。S的(歐氏最近點)Voronoi圖形定義了一個分派函數(assignment function)A : X → S , A(x) = VR(s) 在此VR(s)表示址s ∈ S的Voronoi區域

給予S和X, 我們希望能夠藉由變更定義S的Voronoi圖形的距離函數, 去改變分派X到S的結果。為了達到這個目的, 我們賦予了一組權重函數W = {w(s) | s ∈ S}給這些點址, 並且用下面的力量函數, 取代x點和 s址之間的歐氏距離函數d(x , s)
pow_w(x , s) = d(x , s)^2-w(s)。
就如上述, 我們可獲致一個分派函數A_w : X → S ,它取決於所選擇的權重函數W。在d維空間中, 點集合X對於址集合S之點址(或是群集中心)的最小平方叢集( least-squares clustering)被定義為最小化址和它相關點之間的歐氏距離平方和。我們感興趣的是受限(constrained)最小平方叢集, 即每個址之容量(與其相關點之個數)是有限定的。有趣的是, 用力量圖形所定義出來的分派函數, 就會得到受限最小平方叢集。                                                                            
本計劃我們將探討許多幾何上的叢集問題, 透由尋找它們許多有用的性質來發展更有效率的演算法, 並進而試著實作出這些演算法。




Voronoi diagrams partition their underlying space according to a given set of sites and a distance function. In this sense, they naturally induce a clustering of space. Consider a set S of n point sites, and another set X of m points, in d-space R^d. The (closest-site Euclidean) Voronoi diagram of S defines an assignment function A : X → S, given by A(x) = s if and only if x ∈ VR(s). Here VR(s) denotes the Voronoi region of site s ∈ S.
We will discuss the power diagrams, Given S and X, we would like to be able to change the assignment by varying the distance function that underlies the Voronoi diagram of S. We attach a set W = {w(s) | s ∈ S} of weights to the sites, and
replace the Euclidean distance d(x, s) between a point x and a site s by the power function
pow_W (x, s) = d(x, s)^2 − w(s).
As above, we obtain an assignment function A_W : X → S which now depends on the choice of weights. A least-squares clustering of a set X of points with respect to a set S of sites (or cluster centers) in d-space is defined to minimize the total squared Euclidean distance between the sites and their associated points. Our interest is in constrained least-squares clusterings, where the number of associated points per site (i.e., its capacity) is prescribed. Interestingly, assignments defined by power diagrams are constrained least-squares clusterings.
In this project we will study some geometric cluster problems and try to find their properties to develop some efficient algorithms and try to implement these algorithms.
PI個人首頁(PI's Information) :
http://www.iis.sinica.edu.tw/pages/dtlee/

實驗室網址(Research Information) :
http://www.iis.sinica.edu.tw/pages/dtlee/


Email :
kero@iis.sinica.edu.tw
廖純中
Liau, Churn-Jung
應用邏輯

applied logic
符號邏輯在各方面的應用

applications of symbolic logic on all aspects
PI個人首頁(PI's Information) :
http://www.iis.sinica.edu.tw/pages/liaucj/

實驗室網址(Research Information) :
http://www.iis.sinica.edu.tw/pages/liaucj/


Email :

張韻詩
Liu, Jane Win Shih
善用巨量開放資料與互聯網於强化災防應對與社區復原能力

Disaster Resiliency through Big Open Data and Smart Things (DRBoaST)
善用巨量開放資料與互聯網於强化災防應對與社區復原能力計畫(DRBoaST project)建立於中研院永續科學研究計畫OpenISDM之研究成果上,持續發展相關防救災之學術與實務研究。DRBoaST計畫之目的包括研發快速獲取災害資料之工具與方法,並專注於產生關鍵性之防救災資料及資訊,用以大幅減少災害所帶來的風險,進一步研發善用開放資料與互聯網的防救災軟體與服務,讓官方與民間在面對緊急災害時,具備更完善之抗災能力與更有效率之應變機制。
DRBoast計畫由五個子計畫組成,分別為SIDiRC (韌性社區的災害資訊與防治策略) 、RTEIC (地震防護與應變即時資訊雲端系統)、DiSRC (災害情境截取與紀錄編輯技術)、ADiPLE (主動備災的智能生活環境) 與CSAI (群眾外包策略用於蒐集災情資料之學理基礎研究)。 SIDiRC 子計畫專注於產生最新且精細的社區防災資訊,進而實地應用於台灣易受災之鄉村區域。RTEIC子計畫已完成建置一個整合氣象局地震速報與學界即時科學成果的資訊系統,除了會持續改進這套系統並擴充即時科學成果之外,並研討如何用大型地震後蒐集的觀測資料去評估地震後之潛災害風險及預測可能的複合性災害。DiSRC 子計畫專注於研發具有前瞻性的軟體工具,提供使用者可以詳細記錄各種災害情景,進而轉化機器可讀的資料格式以便未來利用。ADiPLE 子計畫之目的在於實現無所不在的主動式緊急應變系統,藉由接收災害有關當局的警報訊息,啟動適當的回應措施,提供降低人員傷亡的功效。CSAI子計畫將完成群眾外包災害資訊蒐集系統,並具體運用在缺乏災害資訊的環境中,以利提升與量化決策品質。
DRBoast計畫之預期成果包括以社區為基礎之防救災資訊雲、即時地震資訊雲、以群眾外包為基礎之災情訊息蒐集系統、主動型智能防災應變系統、災害情境截取與紀錄編輯技術,為能明確展現上述系統之優勢與應用層面,計畫成果也包含相關之基礎學理與關鍵技術之發表。
為了使DRBoaST研究成果可以具體落實,DRBoaST計畫將持續與NCDR和CWB保持在去三年執行OpenISDM計畫之密切合作關係,也與致力於巨量資料處理,物聯網設計、雲端與人工運算應用於災害管理的國際研究機構及計畫合作。在台灣的新合作夥伴包括智邦科技公司、新光保全與台灣大學醫院,以順利推動技術轉移之相關工作。


As it is evident from their annual reports, within the first year, all the subprojects are making good progress in reaching their milestones according to the proposed schedule. The Appendix provides a list of the project's accomplishments to date, including journal and conference publications, presentations at major conferences, prototypes and awards. The project has also provided tens of graduate students and summer interns with opportunities and environment for multi-disciplinary applied research and advanced development. Their names are listed on the cover of this report.
  The project as a whole have been interacting and collaborating in the following three ways:
 Between individual subprojects: The collaboration between subprojects 2 (RTEIC) and 5 (CSAI) throughout the year is a win-win effort. Specifically, Subproject 2 is using the volunteer management system (VMS) and crowdsource map manager (CMM) developed by Subproject 5. This collaboration enables Subproject 2 to have the tools better customized to suit their needs. Subproject 5 has been benefited by feedback from Subproject 2 on features and usability of the modules to guide their efforts in enhancing their prototype tools. This collaboration will surely continue throughout the project duration.
 With partners and targets of technology transfer: Within the first year, the project did not participate in further development of CAP-Taiwan Profile as planned, mostly due to that fact that the profile is sufficiently mature and is now in use in Taiwan. The project has been working with SKS Security (新光保全股份有限公司) and Taiwan University Hospital to plan for pilot studies for assessing the usability and effectives of the indoor positioning system [9] for use in large public buildings and a building/environment information mist for fine-scale, location-specific emergency response instruction delivery [1]. Assessing the systems in real-life environments by targeted users is the first step towards enabling their wide use in the future. The project will continue to seek similar opportunities for such collaboration. The project has also been reaching out for other likely collaborators, including companies such as HEX (瑞德感知), a likely target of technology transfer, and Accton (智邦科技), a possible collaborator in our efforts to make critical components such as location beacons and intelligent gateways better and cheaper.
 Case studies: DRBoaST project proposed to carry out two case studies that aim to tie together results of subprojects through test scenarios and experimentations for assessment and validation purposes. One is the earthquake detection, preparedness and response (EDPR) case study, which is the focus of the project this year and first part of the next year. The other is one involving flood and debris flow, which will be developed in mid-2017. The project has been making steady progress in EDPR. Hoping that there will not be a major earthquake in coming years, the project will work with scenarios from past major earthquakes, including the 1935/04/21 新竹-臺中地震 and the 2016/02/06 高雄美濃地震. The next step is fleshing out test scenarios and specify the requirements of tools and devices, including contents and capabilities of the information sources, that will participate in the scenarios. These tasks will be completed early 2017. All the subprojects will have prototype devices and systems ready for experimentation by that time, and the project will be able start their assessment within the test scenarios.
PI個人首頁(PI's Information) :
http://www.iis.sinica.edu.tw/pages/janeliu/

實驗室網址(Research Information) :
http://openisdm.iis.sinica.edu.tw/


Email :
Lynette@iis.sinica.edu.tw
王建民
Wang, Chien-Min
雲端運算與人智運算

Cloud Computing and Human-Centered Computing
(1) 整合記憶體內資料儲存的雲端計算平台:MapReduce是目前利用雲端計算來處理巨量資料方面,最常用的平行計算模型。然而我們發現有一類雲端應用,雖然非常適合MapReduce模型,但是其執行效能卻非常低落,而且計算規模也有很大的限制。這類應用包括用於基因定序的後綴陣列排序和近來很受重視的演化式計算。我們將對現有的Hadoop平台進行擴充和改進,融合記憶體內資料儲存,提出一個泛用的加強型雲端計算平台,以提升執行效能和規模擴充性。我們也將實作後綴陣列排序以及演化式計算,以驗證我們所提出的架構,對於這兩種應用的執行效能和應用規模有多大的提升。我們相信這樣的雲端計算平台不但對於學術研究有很大的貢獻,還能大幅拓展雲端計算平台的應用。

(2) 人智運算的穿戴運算系統:研究穿戴式電腦及裝置在人智計算中的應用,特別是在社交網路方面的應用。我們計劃中的人智運算系統應具備的三種能力:具有解周遭環境與人們情況的能力,可提供使生活更美好的服務,和透過感官與人類自然地互動。為了實現這三個能力,我們計劃中將從三個研究學科來發展:情境識別,雲服務,以及擴增實境。我們計畫透過先進的系統設計,提供更適合未來人類生活以及具備更友善人機互動的應用程式。藉由研究相關的穿戴式電腦及裝置,開發更佳的人機整合功能,並透過社交網路系統之系統分析,研究並開發穿戴式社交網路系統。我們將著重於友善的使用者界面,以提升使用者經驗為目標,並且提供更適合的情境感知技術與實境服務的增強實境功能。

(1) A MapRedice framework with an In-memory Data Store: MapReduce is a powerful programming model for processing large data sets with a parallel, distributed algorithm on clouds. The Hadoop framework is the most popular implementation of MapReduce and widely adopted in the processing of large datasets. However, our previous experience on suffix array construction with Hadoop shows that it might result in excessive disk usage and access. Therefore, the performance is degraded and the scale of the application is limited. In this project, we aim at efficient and scalable processing of expansive MapReduce (EMR) applications with in-memory data stores. EMR applications, including suffix array construction and evolutionary computation, are a group of applications that have performance and scalability issues with Hadoop. We shall integrate an in-memory data store with Hadoop and propose a MapReduce framework for EMR applications  to enhance their performance and scalability. To validate the benefit of the proposed framework, we shall use suffix array construction and evolutionary computation as our testbed.

(2) Wearable Computing Systems and Applications in Human-Centered Computing: The goal of this project is to investigate the application of wearable computers and devices in Human-Centered Computing, especially those applications on social networks. A human centered computing system should have three abilities: understanding the context of the surrounding area and humans, providing the service that makes the lives better, and interacting with human naturally through perception. To realize these three abilities, we plan to adopt three corresponding research disciplines: context recognition, cloud service, and augmented reality. Wearable computers and social network services will be integrated to build the proposed wearable social network system. The proposed system will provide more convenient and user-friendly human-computer interaction.
PI個人首頁(PI's Information) :
http://www.iis.sinica.edu.tw/pages/cmwang/

實驗室網址(Research Information) :
http://www.iis.sinica.edu.tw/page/research/ComputerSystem.html?lang=zh


Email :
cmwang@iis.sinica.edu.tw
呂及人
Lu, Chi-Jen
機器學習與賽局理論

Machine learning and game theory
在日常生活中,我們時常必須不斷在未知的環境中作決定,並為此付出代價。這可被抽象化為所謂的線上決策問題。這個問題不僅是機器學習領域中的重要問題,在其他領域也有不少應用。我們希望能為此問題設計出好的線上演算法,可以從過去的歷史中學習,而能在未來做出好的決定。我們也希望能為此問題在其他領域,特別是賽局理論,找到更多的應用。

Many situations in daily life require us to make repeated decisions before knowing the resulting outcomes and paying the corresponding prices. This motivates the study of the so-called online decision problem, in which one must iteratively choose an action and then receive some corresponding loss for a number of rounds. It is a fundamental problem in the area of machine learning, and it has surprising applications in several other areas as well. We would like to design better online algorithms which can learn from the past and make better decisions as time goes by. We would also like to find more applications in other areas, especially in the area of game theory.
PI個人首頁(PI's Information) :
http://www.iis.sinica.edu.tw/pages/cjlu/

實驗室網址(Research Information) :
http://www.iis.sinica.edu.tw/pages/cjlu/
https://

Email :
cjlu@iis.sinica.edu.tw
葉彌妍
Yeh, Mi-Yen
巨量資料探勘與深度學習模型應用

Algorithms and applications of data mining and deep learning
本研究計畫可學習到資料探勘與深度學習的相關演算法與效能,並了解到可能的相關應用,例如用深度學習模型從事網路廣告競標出價預測、自動QA問答系統、以及增強式學習應用等。

In this project, we will study the related algorithms and models of data mining and deep learning. We will also study the possible applications such as leveraging deep learning models in online ad bidding price prediction, QA systems, and reinforcement learning applications.
PI個人首頁(PI's Information) :
http://www.iis.sinica.edu.tw/pages/miyen/

實驗室網址(Research Information) :
http://www.iis.sinica.edu.tw/pages/miyen/
https://

Email :

鄭湘筠
Cheng, Hsiang-Yun
適用於新世代記憶體技術與資料密集程式之記憶體系統設計

Memory system designs for emerging technologies and data-intensive applications
可位元存取之非揮發性記憶體(NVM)是極具發展潛力之記憶體技術並可能成為未來記憶體與存儲系統的主流,許多公司開始逐漸推出相關產品,例如 Intel 的Optane Memory,來將非揮發性記憶體應用於電腦系統的記憶體架構中。本實習計劃的目標為探討非揮發性記憶體對電腦系統設計所帶來的新挑戰,並針對資料密集程式之應用情境,設計高效能低耗電之非揮發性記憶體系統。實習生可選擇下列研究主題,或其他相關研究議題。
1. 針對非揮發性記憶體之特性重新設計軟體或演算法 (資料索引, 機器學習等)
2. 利用非揮發性記憶體內部之運算能力,減少資料搬移以加速資料密集程式
3. 設計適用於非揮發性記憶體與存儲系統之資料管理機制以提高效能減少耗電

Byte-addressable non-volatile memories (NVMs) are promising and are likely to become the mainstream in the near future. Products such as Intel’s Optane memory are emerging in the market to demonstrate the usage of such new technologies in the memory hierarchy. Our goal is to tackle the design challenges introduced by NVMs to build energy-efficient memory systems for data-intensive applications. Candidate topics include, but are not limited to, the following:
1. NVM-aware software design (data indexing, machine learning, etc)
2. Exploiting processing-in-NVM capability to accelerate data-intensive applications
3. Energy-efficient data management in NVM-based memory and storage systems
PI個人首頁(PI's Information) :
http://www.citi.sinica.edu.tw/pages/hycheng/

實驗室網址(Research Information) :
http://www.citi.sinica.edu.tw/pages/hycheng/
https://

Email :
hycheng@citi.sinica.edu.tw
陳祝嵩
Chen, Chu-song
多領域深度學習技術開發與整合

Multi-modal deep learning technology development and integration
未來AI與深度學習勢必走向跨領域的整合,例如電腦視覺、電腦聽覺、與自然語言處理的深度模型整合與應用。本研究將致力於跨領域的AI系統整合,並培養這方面的人才。

AI and deep learning need the integration of different modalities (such as image, speech, natural language) so that a strong AI system can be built. We hope to hire people who are interested in conducting advanced multi-modal deep learning systems and applications in the related fields.
PI個人首頁(PI's Information) :
http://www.iis.sinica.edu.tw/pages/song/

實驗室網址(Research Information) :
http://www.iis.sinica.edu.tw/pages/song/publications_en.html
https://

Email :
song@iis.sinica.edu.tw
馬偉雲
Ma, Wei-Yun
透過深度學習進行:推薦功能對話機器人/智慧黑白講對話機器人/自動知識學習系統

透過深度學習進行:推薦功能對話機器人/智慧黑白講對話機器人/自動知識學習系統
今年的實習我們邀請同學透過深度學習(Deep Learning)進行以下三項專案的其中之一:

1. 推薦功能對話機器人:透過深度學習當中的增強式學習(Reinforcement Learning),打造一個具推薦功能的機器人,跟一般的FAQ或是單純搜尋任務(如訂票)不同,機器人必須根據對話了解使用者的需求,綜合分析產品或服務,然後以具有邏輯與說服力的說明,來推薦其中的產品或服務,舉個極端例子來說,就算user已經要購買某項特定服務或商品(如買某電影的票),機器人仍然可以根據跟user的對話過程,與其他電影的了解,評估可能還有user更有興趣的電影可以加以推薦。簡言之,我們要打造的不是 “AI服務員”,我們要打造 “AI顧問”. 去年(2017)暑期intern在此項目有豐碩成果,所開發的美妝保養顧問-PerMu機器人拿到2017 痞克邦 HACKATHON 最佳產品力獎 (報導在https://www.youtube.com/watch?v=w7eUtJ_PLU0),目前正與痞克邦洽談合作. 另外,我們也跟LINE Taiwan正進行新聞聊天機器人的合作,將LINE Today的新聞推薦給使用者。

2. 智慧黑白講對話機器人:第二個專案會著重在所謂 chitchat,或稱智慧黑白講,也就是沒有特定目的的聊天. 我們會提供微博抓下來的對話聊天語料作為訓練資料. 目前大多數的作法是深度學習當中的seqence-to-sequence model,我們想將更豐富的語義訊息放在model中,打造新一代semantics-to-sequence model,我們實驗室過去開發了人類的知識網(Ehownet),對semantics的向量化表達有豐富經驗,可以提供intern創新的深度學習探索空間,而不僅僅是implement文獻上已有的作法。

3. 自動知識學習系統:我們知道新的知識會夜以繼日的不斷產生,一個具有AI能力的系統最重要的功能之一就是能夠從大量的資料當中,分析資料,加以理解,組織成結構化知識。我們實驗室過去已經開發了人類的知識網(E-HowNet),打下堅實基礎,此專案的目標是進一步加以擴張,利用機器閱讀的深度學習技術(RNN/CNN)將關鍵的關係三元組合從閱讀的文章中自動抽取出來,如 (”林書豪” ,MemberOf,”籃網隊”) 或是 (“麥特載蒙”,PlayerOf,”心靈捕手”)等等。


今年的實習我們邀請同學透過深度學習(Deep Learning)進行以下三項專案的其中之一:

1. 推薦功能對話機器人:透過深度學習當中的增強式學習(Reinforcement Learning),打造一個具推薦功能的機器人,跟一般的FAQ或是單純搜尋任務(如訂票)不同,機器人必須根據對話了解使用者的需求,綜合分析產品或服務,然後以具有邏輯與說服力的說明,來推薦其中的產品或服務,舉個極端例子來說,就算user已經要購買某項特定服務或商品(如買某電影的票),機器人仍然可以根據跟user的對話過程,與其他電影的了解,評估可能還有user更有興趣的電影可以加以推薦。簡言之,我們要打造的不是 “AI服務員”,我們要打造 “AI顧問”. 去年(2017)暑期intern在此項目有豐碩成果,所開發的美妝保養顧問-PerMu機器人拿到2017 痞克邦 HACKATHON 最佳產品力獎 (報導在https://www.youtube.com/watch?v=w7eUtJ_PLU0),目前正與痞克邦洽談合作. 另外,我們也跟LINE Taiwan正進行新聞聊天機器人的合作,將LINE Today的新聞推薦給使用者。

2. 智慧黑白講對話機器人:第二個專案會著重在所謂 chitchat,或稱智慧黑白講,也就是沒有特定目的的聊天. 我們會提供微博抓下來的對話聊天語料作為訓練資料. 目前大多數的作法是深度學習當中的seqence-to-sequence model,我們想將更豐富的語義訊息放在model中,打造新一代semantics-to-sequence model,我們實驗室過去開發了人類的知識網(Ehownet),對semantics的向量化表達有豐富經驗,可以提供intern創新的深度學習探索空間,而不僅僅是implement文獻上已有的作法。

3. 自動知識學習系統:我們知道新的知識會夜以繼日的不斷產生,一個具有AI能力的系統最重要的功能之一就是能夠從大量的資料當中,分析資料,加以理解,組織成結構化知識。我們實驗室過去已經開發了人類的知識網(E-HowNet),打下堅實基礎,此專案的目標是進一步加以擴張,利用機器閱讀的深度學習技術(RNN/CNN)將關鍵的關係三元組合從閱讀的文章中自動抽取出來,如 (”林書豪” ,MemberOf,”籃網隊”) 或是 (“麥特載蒙”,PlayerOf,”心靈捕手”)等等。
PI個人首頁(PI's Information) :
http://www.iis.sinica.edu.tw/pages/ma/index_zh.html

實驗室網址(Research Information) :
http://ckip.iis.sinica.edu.tw/CKIP/index.htm
https://

Email :
ma@iis.sinica.edu.tw
張原豪
Chang, Yuan-Hao
嵌入式系統之儲存系統介紹

Embedded Systems
前瞻記憶體儲存系統及其應用
嵌入式系統及其作業系統設計


Memory/Storage Systems
Non-volatile Memory
Operating Systems
Embedded Systems
PI個人首頁(PI's Information) :
http://www.iis.sinica.edu.tw/pages/johnson/

實驗室網址(Research Information) :
http://www.iis.sinica.edu.tw/~johnson/index_c.php
https://

Email :
johnson@iis.sinica.edu.tw
莊庭瑞
Chuang, Tyng-Ruey
敏感資料群組共享模式之研究

Communal Sharing of Sensitive Data
巨量資料科技的發展為當代人口、社會行為以及公共衛生等仰賴對巨量資料進行二次利用之系統性研究帶來變革。然而,這些技術性的革命也同時帶來諸多適法性的爭議,例如:資料共享的正當法律程序、隱私權的維護以及個人資料的保護等。除此之外,如何在鼓勵資料共享的同時,透過對資訊流動的妥適管制確保資訊的私隱性,亦屬巨量資料科技下另一重要的倫理道德挑戰。而這些巨量資料科技下的適法性與倫理道德考量,或可透過促進公眾參與之方式,進而將如何衡平資料二次使用所帶來之風險與易用性,以制度化之模式達到對隱私權的保障。

本研究計畫預計提出一可行性架構,並將隱私權納入使該架構以系統化之方式符合法律規範與要求,包括正當法律程序、資料掌控之透明化、社會參與以及負責任的自我管理等。為了改良現有的隱私權框架,本計畫將進一步探討三個研究主題,包括:集結個人資料之規範、基準與管制;共有資料分享之具可保密性以及具可審計性;及以參與者為中心之資料分享管理架構。最終,本計畫期能建構適用於各種領域,並以參與者為中心且以社會為基準,亦能符應社會歸責之資料二次使用隱私框架。

The recent information technology revolution has brought new challenges in the legal arena for the due process data sharing, the right to privacy, and personal data protection. How to appropriately manage the flows of information and to encourage data sharing yet keep shared information private remains a challenge. These concerns have moved beyond the traditional privacy frameworks that focus merely on anonymity and de-identification. Instead, it relies on the establishment of a more socially accountable and communicative framework that not only can balance the risk and usability of secondary data usage, but also can institutionalize that demand by improving public participation.

By critically reviewing existing data access models, techniques and practices, this project aims at proposing a doable framework by designing privacy into a comprehensive system that can accommodate the legitimate requirements of community participation, transparent data control, and responsible self-management in the big data era. Specifically, this project will survey and develop the governing principles of a communal approach to personal data management where members of a community pool sensitive information about themselves for mutual benefits and public good.
PI個人首頁(PI's Information) :
http://www.iis.sinica.edu.tw/pages/trc/

實驗室網址(Research Information) :
http://www.iis.sinica.edu.tw/~trc/
https://

Email :
trc@iis.sinica.edu.tw