Artificial Intelligence

Apply machine learning and deep learning

Looking back on 2019

Happy Holidays, and Happy (almost) New Year! This time of year, I like to look back and remember the things we accomplished in a year. I'll recap the main themes of the deep learning blog this year, and ask you what you'd like to see in 2020.

A shout out to my style transfer blog post.


Theme 1: Guest Bloggers

Guest bloggers are my favorite, since I love reading what other people are doing and writing about in deep learning.
  • Oge Marques wrote about deep learning twice this year, once about image augmentation and once about scene classification. His code is easy to follow and available on File Exchange. I really enjoy working with Oge because his depth of knowledge on deep learning, his attention to details, and his desire to educate and empower his audience. If you haven't already read them, check out the these blog posts:
  • Jakob Kather had a paper published here, and then offered to write a blog post to give us insight into his research on deep learning for medical imaging. All about classifying medical images, Jakob gives us practical deep learning code and dealing with very large images in MATLAB.
  • Barath Narayanan wrote about his work on Malaria Detection and showed transfer learning in action. His work is interesting, he likes MATLAB, and I would welcome working with Barath again in the new year! (hint hint).
These are real people doing real deep learning research, which is motivation for me to keep working on new deep learning techniques and trends. I really hope the trend of guest bloggers continues in 2020. You can always get in touch with me if you'd like to be featured on this blog.
From Oge's post on Image Augmentation.

Theme 2: Helpful tricks for deep learning

There were quite a few opportunities to learn tips and tricks for deep learning:
  • Maria wrote about Ensemble Learning, which is so simple I almost didn't believe it was a real thing. This concept is the perfect post on the blog: something that isn't in documentation, but definitely worth highlighting and very simple to execute in MATLAB.
  • I wrote about CAM visualization, and there are many other visualizations in MATLAB for deep learning. Occlusion sensitivity could be a great topic for 2020.
  • Maria also wrote about feature visualization. I enjoy her easy-to-understand approach to advanced topics, and I hope this continues in the new year.

From Maria's post on Feature Visualization.

Theme 3: New deep learning features

"What's new in deep learning" posts continue to be a fan favorite, and this will continue in 2020 with the release of MATLAB R2020a imminent. Highlights include:
  • R2019a examples and R2019b examples: Not only are these good examples, they also highlight applications of deep learning: audio, wireless, reinforcement learning. This seems to be a good way to compile new examples and keep status of what's new in the product. I'll continue this next year!
Now, I would like to ask you all: what did you enjoy in 2019? What would you like to see more or less of? Our schedule for 2020 is not final yet, so there's still plenty of time to make changes and add topics of interest.
As for me, I'll reemerge after the holidays with next excitement and vigor about delivering new deep learning content. This includes: short and simple posts with code, videos and how-to's, and advanced concepts including insight from our deep learning developers.
Finally, you've made it to the end of this post, so I have "gifts" for you! In addition to the blog, we spend lots of time creating content throughout the year for MATLAB and future MATLAB users.
  • -> Reinforcement Learning Ebooks: In exchange for your email address, there are three ebooks highlighting RL concepts and how to incorporate them in your work. We see Reinforcement Learning as a trend that will continue in 2020, so best to get ahead of it and learn the basics now!
  • -> New Machine Learning Onramp. Learn the key concepts of machine learning in a fun, easy to understand, and interactive way. In addition to a Deep Learning Onramp, the machine learning onramp is a free tool to help familiarize you with the basics and then send you off into the world ready to learn more.
  • -> Data Science Things: Heather Gorr (@HeatherGorr) has been prolific creating excellent data science content this year. This includes a thorough video series on our website here, and a Coursera course. I hope to have more opportunities to work with her next year!

Heather teaching me things in the video studio (as usual!)

  That's all folks! Keep the comments coming, and I hope you have a terrific rest of the year!

  • print
  • send email


댓글을 남기려면 링크 를 클릭하여 MathWorks 계정에 로그인하거나 계정을 새로 만드십시오.