๊ธฐ์ ์ ๋ฐฐ๊ฒฝ์ด ์๋ ์ฐฝ์ ์๊ฐ ๊ธฐ์ ์ด ํ์ํ ์ฐฝ์ ์ ํ๋ ค๊ณ ํ ๋ ์ค์ํ ๋ด์ฉ์ ๋ฌด์์ผ๊น์? ์คํํธ์ ์ ํ์ํ ๊ธฐ์ ๋ค๊ณผ, ์ฐฝ์ ์ ๊ณ ๋ฏผํ ๋ฐฉํฅ์ ์๋ดํฉ๋๋ค.
2017๋ 4์ 27์ผ ๊ตฌ๊ธ์บ ํผ์ค ์์ธ์ Campus For Moms ์์ ๋ฐํํ ์ฌ๋ผ์ด๋์ ๋๋ค.
What is important when a founder who does not have a technical background wants to start a business that requires technology? It introduces the technologies necessary for start-up, and directions to worry when starting a business.
This slide is for invited talk of Campus For Moms on April 27, 2017 at Google Campus Seoul.
Machine Learning Model Serving with Backend.AIJeongkyu Shin
ย
๋จธ์ ๋ฌ๋ ๋ชจ๋ธ์ ์๋น์ค ๋จ์์ ์๋นํ๋ ๊ฒ์ ์์ด ๋ง์ด ๊ฐ๋๋ค.
์๋น์ค ๊ณผ์ ์ ํธ๋ฆฌํ๊ฒ ํ๊ธฐ ์ํ์ฌ TensorFlow serving ๋ฑ ์๋น ๊ณผ์ ์ ๋๋ ๋ค์ํ ๋๊ตฌ๋ค์ด ๊ณต๊ฐ๋๊ณ ๊ฐ๋ฐ๋๊ณ ์์ต๋๋ค๋ง, ์ฌ์ ํ ์๋น ๊ณผ์ ์ ๊ท์ฐฎ๊ณ ๋ถํธํฉ๋๋ค. ์ด ์ธ์ ์์๋ Backend.AI ์ TensorFlow serving์ ์ด์ฉํ์ฌ ๊ฐ๋จํ๊ฒ TensorFlow ๋ชจ๋ธ์ ์๋นํ๋ ๋ฒ์ ๋ํด ๋ค๋ฃจ์ด ๋ด ๋๋ค.
Backend.AI ์๋น ๋ชจ๋๋ฅผ ์๊ฐํ๊ณ , ์ฌ๋ฌ TF serving ๋ชจ๋ธ ๋ฑ์ Backend.AI ๋ก ์๋น์คํ๋ ๊ณผ์ ์ ํตํด ์ค์ ๋ก ์ฌ์ฉํ๋ ๋ฒ์ ์์๋ด ๋๋ค.
Serving the machine learning model at the service level is a lot of work. A variety of tools are being developed and released to facilitate the process of serving. TensorFlow serving is the greatest one for serving now, but the docker image baking-based serving process is not easy, not flexible and controllable enough. In this session, I will discuss how to simplify the serving process of TensorFlow models by using Backend.AI and TensorFlow serving.
I will introduce the Backend.AI serving mode (on the trunk but will be official since 1.6). After that, I will demonstrate how to use the Backend.AI serving mode that conveniently provides various TensorFlow models with TensorFlow serving on the fly.
Arduino, Raspberry Pi, Beagleblack and so on, all are signaling new tide of open source hardware.
In other words, open source is widening from software into hardware.
It will also affect the IOT, Internet of Things, as the major IOT frameworks are also open source based.
๊ตฌ๊ธ์ ๋จธ์ ๋ฌ๋ ๋น์ : TPU๋ถํฐ ๋ชจ๋ฐ์ผ๊น์ง (Google I/O Extended Seoul 2017)
์ด ๋ฐํ์์๋ ๊ตฌ๊ธ์ ๋จธ์ ๋ฌ๋ ๋ถ์ผ์ ๋ํ ์ ๊ทผ ๋ถ์ผ, ๋ฐฉ๋ฒ ๋ฐ ๋ชฉํ๋ฅผ ๊ตฌ๊ธ I/O 2017์ ์ธ์ ๋ฐํ๋ค์ ํตํด ์์๋ด ๋๋ค.
From TPU to Mobile: Google's Machine Learning Vision
In this presentation, I will cover about the approaches, methods and goals of Google's machine learning area through the sessions of Google I/O 2017.
Boosting machine learning workflow with TensorFlow 2.0Jeongkyu Shin
ย
TensorFlow 2.0 is the latest release aimed at user convenience, API simplicity, and scalability across multiple platforms. In addition, TensorFlow 2.0, along with a variety of new projects in the TensorFlow ecosystem, TFX, TF-Agent, and TF federated, can help you quickly and easily create a wide variety of machine learning models in more environments. This talk will introduce TensorFlow 2.0 and discusses how to develop and optimize machine learning workflows based on TensorFlow 2.0 and projects within the various TensorFlow ecosystems.
This slide was presented at GDG DevFest Songdo on November 30, 2019.
TensorFlow 2: New Era of Developing Deep Learning ModelsJeongkyu Shin
ย
This document discusses TensorFlow and provides information about:
1. Major TensorFlow releases from 2016-2018 and new features introduced each year like TensorFlow Serving, Keras API, TensorFlow Lite, etc.
2. Common TensorFlow operations that have changed names or behaviors between TensorFlow 1.x and 2.x like tf.variable, tf.global_variables_initializer, and Session.run.
3. How to distribute training across multiple GPUs and CPUs using strategies like MirroredStrategy in TensorFlow 2.x.
์ด ๋ฐํ๋ 2018๋ 4์ 14์ผ ์์ธ์์ ์ด๋ฆฐ TensorFlow Dev Summit Extended Seoul '18 ์์ TensorFlow Dev Summit 2018์ ๋ฐํ ๋ด์ฉ ์ค TensorFlow.Data ๋ฐ TensorFlow.Hub์ ๊ดํ ๋ฐํ๋ค์ ์ ๋ฆฌํ ๋ด์ฉ์ ๋๋ค.
This presentation summarizes the talks about TensorFlow.Data and TensorFlow.Hub among the sessions of TensorFlow Dev Summit 2018, and presented at TensorFlow Dev Summit Extended Seoul '18 held on April 14, 2018 in Seoul.
์ด ๋ฐํ์์๋ TensorFlow์ ์ง๋ 1๋ ์ ๊ฐ๋จํ๊ฒ ๋์๋ณด๊ณ , TensorFlow์ ์ฐจ๊ธฐ ๋ก๋๋งต์ ๋ฐ๋ผ ๊ฐ๋ฐ ๋ฐ ๋์ ๋ ์์ ์ธ ์ฌ๋ฌ ๊ธฐ๋ฅ๋ค์ ์๊ฐํฉ๋๋ค. ๋ํ 2017๋ ๋ฐ 2018๋ ์ ๋จธ์ ๋ฌ๋ ํ๋ ์์ํฌ ๊ฐ๋ฐ ํธ๋ ๋์ ๋ฐฉํฅ์ ๋ํ ์ด์ผ๊ธฐ๋ ํจ๊ป ํฉ๋๋ค.
In this talk, I look back the TensorFlow development over the past year. Then discusses the overall development direction of machine learning frameworks, with an introduction to features that will be added to TensorFlow later on.
Deep-learning based Language Understanding and Emotion extractionsJeongkyu Shin
ย
This document discusses natural language understanding and emotion extraction using deep learning. It summarizes SyntaxNet and DRAGNN, which are frameworks for natural language processing using deep learning. It then discusses using these frameworks to build models for part-of-speech tagging, dependency parsing, and language understanding. It also discusses building models for extracting emotions from text using techniques like SentiWordNet and modeling emotions in a vector space.
OSS SW Basics Lecture 03: Fundamental parts of open-source projectsJeongkyu Shin
ย
This document discusses the fundamental parts of open-source projects, including common development positions and responsibilities, types of projects, important tools for collaboration and testing, and best practices for documentation. It also provides an overview of a lecture on these topics and gives an assignment for students to explore 10 related GitHub projects and summarize the results.
OSS SW Basics Lecture 02: History, culture and community of open-sourceJeongkyu Shin
ย
This document provides a summary of the history and culture of open-source software. It discusses how early software in the 1950s-1960s was mostly open-source and shared within research communities. Major developments included the GNU project in 1983, Linux in 1991, and the establishment of the Open Source Initiative in 1998. The document also contrasts the philosophies of the free software movement promoted by the Free Software Foundation, which focuses on licensing and ethics, versus the open source movement championed by the Open Source Initiative, which is more enterprise-friendly and pragmatic. Community has always been at the core of open source through user groups, developer communities, and more.
This document summarizes the syllabus and schedule for a 16-week open source software boot camp. It includes details about assignments, exams, attendance policy, and topics to be covered each week. The course will involve 1 hour of lecture and 1.5 hours of lab each week. There will be one midterm exam and a final project. Attendance is mandatory for at least 11 lectures and absences will result in a grade reduction. Students are asked to form project teams based on their interests and skills.
The bleeding edge of machine learning stream in 2017 - APAC ML/DS Community ...Jeongkyu Shin
ย
Video (Korean): https://www.youtube.com/watch?v=r64_PeoZvao
๊ธฐ๊ณํ์ต์ ์ต๊ทผ์ ์ฐ๊ตฌ ์ฑ๊ณผ ๋ฐ ๊ธฐ์ ์ ๋ฐ์ ์ ํ์ ์ด ๋ค์ํ ๋ถ์ผ์ ๋ณธ๊ฒฉ์ ์ผ๋ก ์ ์ฉ๋๊ธฐ ์์ํ์ต๋๋ค. 2017๋ ์ ์์ฉ๋ถ์ผ์ ํ์ฅ์ ํ์ ์ด ๊ธฐ๊ณํ์ต ์์ฉ์ด ๋์คํ๋๋ ํ ํด๊ฐ ๋ ๊ฒ์ ๋๋ค. ์ด ๋ฐํ์์๋ ๊ธฐ๊ณํ์ต์ด ํด๊ฒฐํ ๊ธฐ์ ์ ์ธ ๋ฌธ์ ์, ํ์ฌ ํด๊ฒฐํ๋ ค๊ณ ํ๋ ๋์ ๋ค์ ๋ค๋ฃน๋๋ค. ๋ํ 2017๋ ํ์ฌ ๊ธฐ๊ณํ์ต์ด ์์ฉ๋๊ณ ์๋ ๋ถ์ผ๋ค๊ณผ ์์ฉ ๋ฐฉ๋ฒ ๋ฐ, ์ดํ ๊ธฐ๊ณํ์ต ์ ์ฉ์ ํตํด ๋ฐ์ ํ ์ ์๋ ๋ถ์ผ๋ค๊ณผ ์ ์ฉ ์์ด๋์ด๋ฅผ ์ด์ผ๊ธฐํฉ๋๋ค.
Machine learning has been applied to various areas in earnest owing to recent research results and technological advancements. In 2017, machine learning application will be popular with the expansion of the application area. This talk covers technical issues solved by machine learning, and difficult problems that should be solved now. It also covers the areas that apply machine learning in 2017, application methods, area that can develop by application machine learning, and application ideas.