Welcome to Machine Learning Up-To-Date (ML UTD) 3! The LifeWithData blog separates the signal from the noise in today’s hectic front lines of the intersection between software engineering and machine learning.
LifeWithData aims to consistently deliver curated machine learning newsletters that point the reader to key developments without massive amounts of backstory for each. This enables frequent, concise updates across the industry without overloading readers with information.
ML UTD 3 brings innovations in the areas of data, computing, and platforms.
[Data] Look, High Dimensions!
Data visualization is all fun and games until you try to use it on a vector in high dimensions. As an alternative to TSNE or UMAP, Facebook built HiPlot, which uses parallel plots to discover correlations and patterns in high-dimensional data. The library seems great; I’ll likely do a detailed write-up on this one. Check out their repository here.
[Data] Data Begets Much More Data
A new version of the fantastic imgaug image augmentation library is out. Data augmentation is a handy tool for helping data-hungry neural networks generalize to new data.
Naver Labs Europe has released the KITTI2 data set. I’m sorry to be the bearer of bad news, but it’s not a data set of images of kitties. However, it’s still pretty awesome, and even more useful than pictures of cats. The data set is a collection of photo-realistic video footage from the Unity game engine. This kind of data helps in leaps and bounds to train neural networks for autonomous vehicle navigation.
[Computing] Colab, Take My Money!
In what computing environment do you perform your exploratory data analysis? Vanilla Jupyter notebooks are so “yesterday.” Google’s Colab environment offers a more batteries-include version. Now, Colab has introduced a paid offering. For a monthly cost of $10, your notebooks will have a longer runtime, more RAM, and access to more powerful GPUs. Read more on their signup page.
[Computing] Data Science, Not DevOps
Speaking of evolutions of the Jupyter notebook, DeepNote has raised funding to marry the best of the traditional IDE with the Jupyter notebook. I’m pulling for them!
[Computing] Oh My Bash with AI
What’s your favorite terminal? I bet it isn’t powered by AI…yet. IBM has developed the open-source CLAI to leverage natural language processing to make your life on the bash terminal even more glamorous. Its current skills include the following:
- Natural language to command (nlc2cmd): from a natural language question about the desired command, suggest the correct terminal syntax for it
- Cloud deployment automation: learn how to perform cloud software deployments, based on historical deployments performed manually
- Fixit: from a failed command, suggest the appropriate set of commands that will accomplish what CLAI believes the user intended
- Man page explorer: from a natural language description of a desired command, fetch and open the “man page” (manual) for the command
[Platforms] Tensorflow.js Hits React Native
Stay Up To Date
That’s all for ML UTD 3. However, things are happening very quickly in academics and industry! Aside from ML UTD, keep yourself updated with the blog at LifeWithData.
If you’re not a fan of newsletters, but still want to stay in the loop, consider adding lifewithdata.org/blog to a Feedly aggregation setup.