How to Transform Research Oriented Code into Machine Learning APIs with Python – PyCon Taiwan 2019
Day 1, R0 13:15–13:45
Recently, web application engineers have more opportunities to work with data scientists than before. At the same time, they are often faced with the research-oriented code written by researchers or data scientists. In order to integrate this code with systems such as APIs, web application engineers need to additionally write the code or refactor it.
This talk covers the gap between the research-oriented code and production code of machine learning API. What is the gap between them? How it can be implemented based on real-world python code? How can the code validate whether the dataset is correct? How can machine learning models be continuously inspected? Audiences can earn the answers to these questions from this talk.
Jesse is a software engineer working at Classi, which is the leading EdTech company in Tokyo. He has been developing a recommender engine which can optimize appropriate learning materials according to learner's abilities. It is based on a statistical method well-known in educational psychology and implemented with python. Prior to this, he used to research the relationships between online learning behaviors and learning outcomes at the UCL Institute of Education (IOE) in the UK. His interest is in how to bridge the gap between data science and engineering. ... https://www.youtube.com/watch?v=9o4f7SXaEJk
Day 2, R3 11:10–11:40
Constantly waiting for your data processing code to finish executing? Through real-life stories, we will explore how to leverage on parallel and asynchronous programming in Python to speed up your data processing pipelines - so that you could focus more on getting value out of your data. While this talk assumes a basic understanding of processes in data pipelines and data science workflows, anyone with a basic understanding of the Python language would be able to understand the concepts and use cases illustrated with analogies.
Slides: https://speakerdeck.com/ongchinhwee/speed-up-your-data-processing-parallel-and-asynchronous-programming-in-python
Speaker: Chin Hwee Ong
Ong Chin Hwee is a data engineer, aspiring polymath and Industry 4.0 enthusiast who happens to be interested in things that fly (and stuff that burn to keep things flying). Hailing from a background in aerospace engineering and computational modelling, Chin Hwee has experience working on innovative projects in collaboration with academia and industry partners. Chin Hwee is a contributor to the documentation for pandas 1.0 and enjoys sharing her experiences at meetups and conferences.
...
https://www.youtube.com/watch?v=w2eUdxPQQ78
Day 2, 13:50–14:20
Zendesk's Chat backend platform is largely powered by Python. Over the years, we have introduced hundreds of open source libraries into our codebase. We have seen multiple production incidents that arise from improper pinning of package versions or incompatible versions between different libraries. Thus a proper application management resolution solution is in need.
Firstly, I will introduce the concept of application dependency management and why it is a hard problem.
Then I will go in depth to share the journey of how we in Zendesk resolved the dependencies issue and slowly prepare itself for the new pip resolver. In particular, I will talk about the behavior of different pip versions when it comes to handling version conflicts and what pip check is and how to use it.
Lastly, I will briefly mention other tools available and give recommendation when to use which and answer questions from the audience.
Slides: https://drive.google.com/file/d/1AZoWKI3OQfpFETD2NoT6C9spanrsrOSG/view?usp=sharing
Speaker: Liuyang Wan
to be added
...
https://www.youtube.com/watch?v=DGL7SYZk_Oc
PyCon Taiwan 2016|一般演講 Talks
? 摘要 Abstract ?
Jupyter (formerly a part of IPython) is a great application for interactive data science and scientific computing, especially the Jupyter Notebook. The Jupyter kernel makes it possible to run Jupyter Notebooks (or console and qtconsole) with other programming languages instead of Python. This talk is going to be an explanation of how it's working under the hood, and how you can create your own Jupyter kernel.
? 關於講者 About Speaker - 廖偉涵 Adrian Liaw ?
高中三年都在家自學還可以延畢的高三自學生,用行動證明就連缺乏教育的屁孩也可以學會寫 Python,常常被嘴說現在的小朋友都怎樣怎樣的。
Adrian is a guy who doesn't really know what he's doing and seriously needs some life advises. He's currently a high school student but not actually in a physical high school because of the lack of school that accepts him. Oh by the way he does Python.
#python #pycontw #pycontw2016
Follow “PyCon Taiwan”
⭐️ Official Website: https://tw.pycon.org
⭐️ Facebook: https://www.facebook.com/pycontw
⭐️ Instagram: https://www.instagram.com/pycontw
⭐️ Twitter: https://twitter.com/PyConTW
⭐️ LinkedIn: https://www.linkedin.com/company/pycontw
⭐️ Blogger: https://pycontw.blogspot.com
...
https://www.youtube.com/watch?v=w1eev9-hqzE
Day 2, 15:10–15:40
在資訊量爆炸的現在,任何媒體甚至是電商都需要一個能夠過濾雜訊,將有效的資訊正確傳遞給各個不同用戶的推薦系統。標題中的 modern 指的是一個包含召回、排序兩階段推薦的現代化作法,而 large content 則是要談如何透過 pyspark 建立可規模化的推薦系統框架。最後要和大家分享,一個擁有數億篇文章與數千萬每日瀏覽量的部落格網站如何從無到有建立這一套作法。
Slides: https://www.slideshare.net/CyrusChienChingChiu/modern-recommender-system-in-large-content-website
Speaker: Cyrus Chiu
喜歡影像創作,從獨立製片工作轉換跑道到電腦視覺,現在認為寫程式也是一種創作。曾任PIXNET演算法工程師,負責推薦系統的主要開發。現為Vpon資料科學家。
...
https://www.youtube.com/watch?v=8xm4L-ecu8c
PyCon APAC 2022|一般演講 Talks|國泰金控 Cathay Financial Holdings / 美光科技 Micron 冠名贊助
✏️ 共筆 Note:https://hackmd.io/@pycontw/r1O1wpmyo
?? Slido:https://app.sli.do/event/mkwQmSfgPv8JCsh8Jny9xm
? 語言 Language:英文 English
? 層級 Level:中階 Intermediate
? 分類 Category:科學 Science
? 摘要 Abstract ?
Scanning Tunneling Microscope is widely used for 2D material characterization and even more, for building quantum computer at the cutting edge of science. A typical surface science experiment can generate hundreds of STM images each containing multiple structures to be identified. Since each investigation is different, it needs to draw on highly flexible methods algorithmically and integrate several image processing techniques. In this talk I’ll present an end-to-end Pythonic solution: reading raw STM data, implementing various STM artifact, noise filtering schemes and flattening. Then I’ll discuss how to search for, identify, count and measure different surface structures, particularly some epitaxial islands. Finally I’ll show statistical results of size and angular distributions, and apply some similar image processing, feature recognition and image interpolation algorithms to extract angular distributions from electron diffraction measurements and compare the results.
? 說明 Description ?
Scanning Tunneling Microscope (STM) is a kind of microscope used to resolve surface structures under sub-nano scale such as features of sizes, shapes and orientations and their electronic band structures as well. It also allows for precise atomic manipulation based on which creation of quantum bits is achieved and propels building of a quantum computer. STM tip scanning is regularly affected by thermal drift, atoms getting picked up by, and moving around on the tip. I’ll use both straightforward NumPy and where possible introduce STM analysis libraries to show how these effects can be mitigated. Once that is done, I’ll first illustrate how edge detection and alignment with respect to surface atoms can be implemented directly in NumPy in various ways, then introduce OpenCV. Since the shapes can merge partially as they grow, I’ll describe how overlapping shapes can still be distinguished and measured. While OpenCV is powerful, it is necessary to do a significant amount of image pre-processing before reliable and repeatable shape identification is working smoothly. Finally, I'll move to Low Energy Electron Diffraction (LEED), which is also quite commonly used for determination of surface structure. LEED images can basically be viewed as 2D Fourier-transformed images of their real-space counterparts (STM images here). After showing how to use numpy's map coordinate to get real space coordinates, I'll show how to fit Gaussian spots and arcs, find them on the screen and once again get distributions from them.
? 講者介紹 About Speaker - Hsu-Kai Cheng ?
I am currently a research assistant at Center for Condensed Matter Sciences in National Taiwan University (NTU). I received my bachelor's degree in the Department of Physics, NTU and my master’s degree in the Graduate Institute of Applied Physics, NTU. My scientific interests include surface science device development and characterization and my current research focuses on post-graphene materials such as black phosphorus and transition metal dichalcogenides. Especially, I aim to explore the properties of topological materials in which the interplay between topology and functions is so overwhelming and renders materials of this kind really promising for future application.
#pycontw #pyconapac2022 #python #scanning #tunneling #microscopy #imageprocessing #featurerecognition #featureextraction
Follow “PyCon Taiwan”
⭐️ Official Website: https://tw.pycon.org
⭐️ Facebook: https://www.facebook.com/pycontw
⭐️ Instagram: https://www.instagram.com/pycontw
⭐️ Twitter: https://twitter.com/PyConTW
⭐️ LinkedIn: https://www.linkedin.com/company/pycontw
⭐️ Blogger: https://pycontw.blogspot.com
...
https://www.youtube.com/watch?v=8g6_Q2zMJ0g