Startups are all the rage now. After all, who doesn’t want to do cool stuff, apply cutting edge theories to work, be ones own boss, and then make tons of money along the way? Dr. Albert Y. C. Chen, Chief Scientist / CTO at Viscovery and a NTHU alumnus, will share his years of tears and cheers as a serial early employee at a number of startups in Taiwan and the United States. Dr. Chen will talk about his struggles of making Computer Vision algorithms “work”, covering a wide range of topics from Intrinsic Imaging, Video Compression, Autonomous Vehicles, Structure From Motion, Feature Matching based Visual Search, as well as Deep Learning based Image Classification, Object Detection, to Video Summarization.

Dr. Albert Y. C. Chen is currently the Chief Scientist / CTO at Viscovery, the leading visual search provider in the Asia Pacific region. Chen was Principal Scientist at Nervve Technologies, the leading provider of interactive visual search tools for the US defense and intelligence community. Previously, Chen worked full time at Tandent Vision Science, GE Global Research, Siemens, and consulted for a number of Computer Vision startups. Chen obtained his Ph.D. at SUNY Buffalo, his M.S. from NTNU, and his B.S. from NTHU, all in Computer Science. Chen have TA’ed/taught more than 10 courses and inspired more than 30 M.S./Ph.D. students into the Computer Vision field during his graduate career. Chen is the recipient of the prestigious Excellence in Teaching Award from the State University of New York.

With more and more Drosophila brain images generated, it is an important work to understand the image similarities, signal transduction and functions, and interaction networks among them. However, it is difficult and time-consuming for processing a large amount of images with large sizes. Therefore, new IT technologies are needed to compress, compare, compute, and visualize these images under the analysis requirements. Recently, we have designed an analysis platform for Drosophila brain images based on the multi-core CPUs and GPUs environment. These images are compressed and then the similarities among them are calculated. These generated images are further re-constructed by the visualization technologies with various types. Finally, the complex and complete interaction networks are constructed according to their images and functions.

Chun-Yuan Lin was born in 1977. He received a B.S. degree in Department of Information Engineering and Computer Science from Feng Chia University in 1999, and the M.S. and Ph.D. degrees in Department of Information Engineering and Computer Science from Feng Chia University in 2000 and 2003, respectively. He joined the Institute of Molecular and Cellular Biology and the Department of Computer Science at National Tsing Hua University as a post-doctoral fellow in 2003 and 2006, respectively. In 2007, he joined the Department of Computer Science and Information Engineering at Chang Gung University as an assistant professor, and became an associate professor in 2013. He also is a faculty member in Research center for Emerging Viral Infections at Chang Gung University from 2007. He served as the project manager in 10 research projects funded by the Ministry of Science Technology, Chang Gung University and Chang Gung Medical Foundation. He also served as the project manager in 4 Industry-University projects funded by ITRI, NVIDIA, Sparkle, and HONGHUTECH. He served as the Referee in more than 50 International Journals and Conferences. He also served as editor/guest editor/chair/program committee in more than 20 International Journals and Conferences. He has 49 papers in journals/books and 62 articles in proceedings of conference/workshop. His research interests are in the areas of parallel and distributed computing, parallel algorithms, bioinformatics, computational chemistry, and neuroinformatics.

你的夢想有多大？你有嘗試錯誤的勇氣嗎？
五個人、一間波士頓老公寓，一個用人工智慧改變世界的夢想正在這裡孵育！
甫獲得國際頂尖創投約10億元新臺幣資金挹注的Appier
以人工智慧為基礎，幫助企業解決跨螢世代中的各種艱鉅挑戰。
這個一開始由五人在客廳創業的團隊究竟是如何吸引到全球創投的注意？

Appier執行長暨創辦人游直翰將與你分享他從學界走向業界的創業歷程、以及在學期間該如何累積實力與競爭力；
在面臨抉擇時，又該如何堅持夢想，走出一條自己的康莊大道！

With developments in handsets equipped with enhanced computational power and expanded communication capabilities, current mobile devices are capable of　taking photographs and transferring images/videos to the Internet. The remarkable growth of smartphones and digital cameras illustrates the　urgent　need for advanced photo capturing and manipulating techniques. This talk　will　include two recent works on aesthetic quality assessment in images and　videos,　and share our experiences in participating ACM Multimedia Grand Challenge　2012 and 2014.

Mei-Chen Yeh is an associate professor in the Department of Computer Science and Information Engineering at National Taiwan Normal University. Her research interests include pattern recognition and multimedia computing. She received her PhD in computer engineering from the University of California, Santa Barbara, where she was a member of the Learning-Based Multimedia Lab. She is a member of IEEE. Contact her at myeh@csie.ntnu.edu.tw.

The ultra low-latency operations of communications and computing enable many potential IoT applications and thus have gained widespread attention recently. Existing mobile devices and telecommunication systems may not be able to provide the highly desired low-latency computing and communications services. To meet the needs of those applications, we introduce Fog-Radio Access Network (F-RAN) architecture which brings the efficient computing capability of the cloud to the edge of network. In this talk, we first introduce the F-RAN and its rationale in serving ultra low-latency applications; then we depict the need of a service framework for F-RAN to cope with the complex tradeoff among the performance, the computing cost, and communication cost. We will also present some other technical challenges and open issues for deploying F-RAN in future mobile telecommunications networks. Short Bio: Ai-Chun Pang received the B.S., M.S. and Ph.D. degrees in Computer Science and Information Engineering from National Chiao Tung University, Taiwan, in 1996, 1998 and 2002, respectively. She joined the Department of Computer Science and Information Engineering (CSIE), National Taiwan University (NTU), Taipei, Taiwan in 2002. She is now the Director of Graduate Institute of Networking and Multimedia (INM), and a Professor in CSIE and INM. She is also an Adjunct Research Fellow of Research Center for Information Technology Innovation, Academia Sinica, Taiwan. Her research interests include the design and analysis of wireless and multimedia networking, mobile communications, and cloud datacenter networking.

人生永遠往下一個階段前行。每逢春夏交際時節，畢業生的心情總會興奮又帶點忐忑不安。清、交大工程學系的畢業生，有很大比例會加入電子資訊產業，成為台灣產業界的中堅份子。本次討論內容將分享從一個『學生』轉變為『工程師』所會面臨的問題以及心態上的調整。希望能讓各位無縫接軌，順利邁入下一個階段。

Open source system software has become the trend for the information technology industry as the complexity of software and the speed of software development have outgrown traditional models where software development and licensing are done under the control of private companies. Thus, not only Linux is being widely used in smartphones and data centers, more and more heavy weight software are released as open source. For example, many Apache open source projects are used for big data analytics, such as Hadoop and Spark. At the same time, big companies such as Google and Microsoft have open-sourced their machine learning software. However, as the open source trend enables developers to reduce the costs for systems and services, how do companies differentiate their products to compete in the market? This talk provides some thoughts and tries to provoke some discussions on this important question.

Shih-Hao Hung is currently an associate professor in the Department of Computer Science and Information Engineering at National Taiwan University. His research interests include parallel processing, heterogeneous computing, hardware-software co-design, performance tools, and virtual prototyping. Shih-Hao has been working with top companies, including IBM, AMD, NVIDIA, Oracle, MediaTek in optimizing application performance on the latest IoT, cloud, and big data systems with heterogeneous multicore architectures, GPGPU and FPGA. He worked on high-performance server systems for Sun Microsystem Inc. in California (2000-2005). He received Ph.D. and M.S. degrees from the University of Michigan, Ann Arbor in 1998 and 1994, respectively. He graduated from National Taiwan University with a BS degree in Electrical Engineering in 1989.

丁俊宏以其系統和資訊專業背景，結合行動裝置和雲端應用服務，創辦第五代虛擬科技公司 VMFive，集結雲端化、虛擬化、App串流、大數據等技術，打造全球最創新行動廣告體驗技術「AdPlay」，讓使用者在瀏覽器中「先體驗，後下載」，此模式讓App廣告轉換率大幅提升，再利用各種資料分析技術來處理數位廣告系統累積的資訊，進而創造各種新的行銷方法。 VMFive因堅強的技術實力與新穎的商業模式，接連在國內外創業競賽大放異彩，包括獲得 2014年 FITI 創業競賽首獎 200 萬、2014年北京 Techcrunch 雙料冠軍、2015年日本 Slush Asia 創業大賽冠軍等。 2015年12月 VMFive 發佈公司 A 輪募資成果，募資金額達 600 萬美金，投資者包含開發國際、趨勢科技、通用移動、心元資本等，並於2016年初在日本設立子公司。

In addition to being applied widely in server and desktop systems, multi-processor technology has received a great interest in embedded application domains. While porting multi-threaded embedded software to a multicore processor with SMP-ready RTOS seems straightforward, developers may be cautious that synchronization errors that are benign on a uniprocessor system may surface in multiprocessor execution. Furthermore, the non-deterministic behavior of multi-threaded embedded software makes dynamic analysis, cyclic debugging, and program instrumentation difficult. Any instrumentation overhead for gathering execution information may change the execution of a program, and lead to distorted analysis results, i.e., probe effect. In this talk, we will present a framework that tackles the non-determinism and probe effect incurred in dynamic analysis of embedded software. We will first discuss a deterministic replay framework to provide reproducible execution. Once program execution events are recorded, their happens-before relation is analyzed and used for replay operations. Software instrumentation and profiling can be safely applied without any probe effect. Then, a discussion of probe effect is presented and a simulation-based analysis is proposed. The execution model of a program is constructed from the trace of instrumented program execution and is used in a simulation analysis where instrumentation overhead is removed. As a consequence, we can infer the ordering of events in the original program execution and verify the existence of probe effect resulted from instrumentation.