Interactive Viz² and Differentiable Machine Learning – Friday Faves

This Friday we’ll have a look at two interactive data viewers and a differentiable ML.

Nd Array Viewer

Napari is an interactive Nd array Viewer with support for fast inspection and analysis of multidimensional data. Definitely something to pop out in a Jupyter notebook and take a look at the processed data.

The Blog Post with even more features is here: https://ilovesymposia.com/2019/10/24/introducing-napari-a-fast-n-dimensional-image-viewer-in-python/

Find the Github repo here: https://github.com/napari/napari
or jump in and

pip install napari

Physics-based Machine Learning Framework

DiffTaichi provides a framework for physics simulation in a differential way. Whether you’d like to tack your simulator on to a CNN to generate samples on the fly, or spice up your GAN-game, this is probably a tool, you’ll want in your toolkit. I have been quoted before that “Physics-based ML is just figuring out how to make your stuff differentiable”.

It’s part of the TaiChi language: https://github.com/yuanming-hu/taichi in the examples – difftaichi.

Check out the cute moving bots and impressive liquid models here:

Exploratory Data Analysis

Apparently SandDance was a thing before, but it’s open-source now and you can integrate it into VSCode.

SandDance by Microsoft

Read the blog here: https://cloudblogs.microsoft.com/opensource/2019/10/10/microsoft-open-sources-sanddance-visual-data-exploration-tool/

Get the VSCode extension here, and make sense of your CSV: https://marketplace.visualstudio.com/items?itemName=msrvida.vscode-sanddance

Bio
Latest Posts

Jesper Dramsch

PhD Student at The Way of the Geophysicist

... is a geophysicist by heart. He works at the intersection of machine learning and geoscience. He is the founder of The Way of the Geophysicist and a deep learning enthusiast. Writing mostly about computational geoscience and interesting bits and pieces relevant to post-grad life.