This article was originally published by AI Frontier.
Kaggle offers four free online courses, including machine learning and deep learning
Clue Editor | Tina
Natalie Debra, Writing editor
“Kaggle, the leading data modeling and analysis competition platform for machine learning and data science, has launched Kaggle Learn, a free online learning program for those who want to Learn machine learning and data science. The course places more emphasis on code practice, and users spend more time writing code themselves than reading it. According to the website, “Users will gain the theoretical knowledge they need to make better modeling decisions, without wasting time on theoretical and historical background that is not conducive to becoming a practical data scientist.” The course includes four courses in machine learning, R language, data visualization and deep learning.”
Kaggle is the world’s largest community of data scientists and machine learning developers, with hundreds of thousands of users, making it an authoritative platform in the industry.
On March 8, 2017, Fei-fei Li, director of Stanford’s ARTIFICIAL Intelligence Lab and chief scientist of Google Cloud, led the case for Google to acquire Kaggle. “Kaggle is the best place to search and analyze public data sets, develop machine learning models and advance data science expertise,” Li said about a year ago. This is a high recognition of Kaggle’s position in machine learning and artificial intelligence, and also makes a reasonable explanation for this year’s acquisition behavior.
For Kaggle, the addition of Google Cloud will give the community better access to and storage of large data sets. Community members will be able to enjoy the most advanced cloud machine learning development environment. This collaboration will certainly be a big boost to the Kaggle community, so Kaggle’s position in ML and AI will only become stronger in the future. (from zhihu user a2Mia elder sister, link: www.zhihu.com/question/32.)
As a machine learning and data science platform for companies and researchers to publish data on Kaggle, the platform’s competitions have attracted statisticians and data mining experts, including a $3 million Heritage Health Prize, Participating in the competition is a learning and practice opportunity for competitors, because only the best will get the prize for the final first place, and the results of the competition will be a very prominent experience on the resume.
Overview of the Kaggle Learn project
The Kaggle Learn program, which is free online, aims to help prospective competitors or data science learners Learn theoretical knowledge before modeling to improve their ability to solve real-world problems, according to the program’s website.
The program consists of four courses, namely machine learning, R language, data visualization and deep learning. Each course is divided into Level1 and Level2 according to the difficulty level, and the knowledge is taught from simple to deep, covering hot fields such as artificial intelligence and data science.
- Machine learning: This course will give you a quick introduction to machine learning, one of the hottest fields in data science.
- R language: This is a language designed specifically for data analysis. This series of courses includes data setup, machine learning, and data visualization.
- Data Visualization: Visualization is the most dynamic technique in data science, representing data sets in visual, beautiful images.
- Deep Learning: Take machine learning one step further by learning how to use TensorFlow in this course. This new skill will surprise you even more.
The program’s four courses feature project-based learning, easy tracking of progress, support from the world’s largest data science community, and the ability to put project experience on your CV.
Kaggle Learn currently has three instructors:
- Dan Becker is a data scientist who has consulted on data science and technology for six Fortune 100 companies and is a code contributor to the Keras Deep Learning library. He holds a doctorate in econometrics. Currently, I am mainly responsible for machine learning and deep learning.
- Racheal Tatman has been an active user and lecturer of R for many years. She has been a lecturer at Software Carpenty and She Codes Now workshops and holds a PhD in linguistics. Currently, I am mainly responsible for R language courses.
- Aleksey Bilogur is a data expert and contributor to the Python open Source project. He has worked for the Office of the Mayor of New York and CUSP at New York University and holds a BACHELOR’s degree in mathematics. Currently, I am mainly responsible for data visualization courses.
Kaggle Learn course details
Machine learning
Machine learning courses are divided into Level1 and Level2. Beginners of machine learning can start from the introductory course and learn the whole process from principles to data uploading, computing environment setting and modeling step by step. Study notes are attached for reference in each class.
Course links:
www.kaggle.com/learn/machi…
Level 1 includes 8 lessons:
- How the model works: Beginners to machine learning first step
- Build your own machine learning project: upload your data and set up your own computing environment
- Filter data with Pandas: Prepare data for modeling
- Run your first model
- Validate model: Test the performance of the model and replace it with another model if necessary
- Underfit, overfit, and model optimization: Adjust the model to improve performance
- Random forest: Use more sophisticated machine learning algorithms
- Submit to the competition: Be proud of what you did, and watch your project progress in the competition
Level 2 consists of seven lessons covering various problems encountered in machine learning, such as dealing with lost data and working with classified data.
R language
Course links:
www.kaggle.com/learn/r
At present, this course is only open for Level 1, consisting of 6 classes:
- Data Science in R Linguistics (Learn the basics of reading data and building machine learning models)
- Manipulating data with Tidyverse, a powerful and widely adopted library, will greatly improve efficiency
- Data visualization with GGploT2: Although there are many data visualization libraries, most experts agree that GGplot2 is the most powerful
- Writing NLP in R: Subject model
- XGBoost (R) machine learning
- Select the best model with complement (model automatic filtering to make machine learning easier and more efficient)
Data visualization
Course links:
www.kaggle.com/learn/data-…
Also, only Level 1 was opened, with a total of 10 classes covering everything from basic knowledge to advanced operations.
Deep learning
Course links:
www.kaggle.com/learn/deep-…
Also, only Level 1 is open, with 6 classes in total, including:
- Introduction to Deep learning and Computer Vision: a brief introduction to the principles of model image processing
By the end of this lesson, you will have an understanding of convolution, the basic building block of deep learning models for computer vision (and many other applications). Then you can learn to use world-class deep learning models.
- Building convolution model
By the end of this course, you’ll understand how convolution works to achieve a level of computer vision that humans can’t.
- Programming with TensorFlow and Keras
After this tutorial, you will be able to use TensorFlow and Keras programming to program one of the most useful models in computer vision.
- The migration study
By the end of this course, you will be able to use transfer learning to build accurate computer vision models based on customer needs in the absence of data.
- Data to enhance
By learning to use data enhancement, you can achieve far more than you could do if you just had the data, and you can build better models.
- Learn more about deep learning
By the end of this lesson, you will know how to use stochastic gradient descent and back propagation to set weights in deep learning models. While these topics are complex, many experts believe they are the most important idea in deep learning.
That’s the introduction to Kaggle’s new free online courses, all of which can be written and run directly on the Kaggle website, without the need to install environments and plug-ins on your own computer. If you are interested in this series of online courses, access to Kaggle website (www.kaggle.com/learn/overv)… Come back and share your learning experience with the AI front!
For more content, you can follow AI Front, ID: AI-front, reply “AI”, “TF”, “big Data” to get AI Front series PDF mini-book and skill Map.