Every two weeks, we find the most interesting data science links from around the web and collect them in Data Science Briefings, the DataMiningApps newsletter. Subscribe now for free if you want to be the first to get up to speed on interesting resources.
- Oil is the New Data
Big Tech is forging a lucrative partnership with Big Oil, building a new carbon cloud that just might kill us all. - How machine learning is revolutionising market intelligence
The business of gathering market-sensitive information is ripe for automation - You watch TV. Your TV watches back.
“In our latest privacy experiment, we tracked how four of the most popular TV brands record everything we watch” - Are Neural Networks About to Reinvent Physics?
“The revolution of machine learning has been greatly exaggerated.” - Seeing Like a Finite State Machine
“China doesn’t have magic AI, they have the same messy, error-prone snake oil that we use.” - AI Update, Late 2019 – Wizards Of Oz
“As time goes, more and more cracks are showing on the self driving car narrative” — and other issues - Can AI Built to ‘Benefit Humanity’ Also Serve the Military?
Microsoft’s $10 billion Pentagon contract puts the independent artificial-intelligence lab OpenAI in an awkward position. - Adobe Photoshop AI Detector
Trust in what we see is increasingly important in a world where image editing has become ubiquitous – fake content is a serious and increasingly pressing issue. - Safety Gym
“We’re releasing Safety Gym, a suite of environments and tools for measuring progress towards reinforcement learning agents that respect safety constraints while training.” - It’s Sony AI vs. Facebook, Google
Sony Corp. has launched Sony AI, a new organization to pursue advanced R&D in artificial intelligence. - Google Cloud Explainable AI
“We are excited to announce our latest step in improving the interpretability of AI with Google Cloud AI Explanations.” Spoiler: it’s mainly Shapley values, but the white paper is an interesting read - The 8 Minute Guide To How Your Business Can Solve Problems with AI and Machine Learning
If you have known categories of something, machine learning can sort things into those categories. - Making Git and Jupyter Notebooks play nice
“jq rocks for speedy JSON mangling. Use it to make powerful git clean filters, e.g. when stripping out unwanted cached-data from Jupyter notebooks.” - Lessons learned building an ML trading system
One of the better posts on the topic of using ML for trading, though still take it with a grain of salt. - roughViz: a JavaScript library for creating sketchy/hand-drawn styled charts in the browser
“Use these charts where the communication goal is to show intent or generality, and not absolute precision. Or just because they’re fun and look weird.” - Wide Feedforward or Recurrent Neural Networks of Any Architecture are Gaussian Processes
“We show that this Neural Network-Gaussian Process correspondence surprisingly extends to all modern feedforward or recurrent neural networks composed of multilayer perceptron, RNNs (e.g. LSTMs, GRUs), (nD or graph) convolution, pooling, skip connection, attention, batch normalization, and/or layer normalization.” - Machine Learning on Encrypted Data Without Decrypting It
“Recent breakthroughs in cryptography have made it practical to perform computation on data without ever decrypting it.” - How to apply machine learning and deep learning methods to audio analysis
Building machine learning models to classify, describe, or generate audio typically concerns modeling tasks where the input data are audio samples. - Understanding the generalization of ‘lottery tickets’ in neural networks
“The lottery ticket hypothesis, initially proposed by researchers Jonathan Frankle and Michael Carbin at MIT, suggests that by training deep neural networks (DNNs) from “lucky” initializations, often referred to as “winning lottery tickets,” we can train networks which are 10-100x smaller with minimal losses” - BodyPix: Real-time Person Segmentation in the Browser with TensorFlow.js
“We are excited to announce the release of BodyPix, an open-source machine learning model which allows for person and body-part segmentation in the browser with TensorFlow.js. With default settings, it estimates and renders person and body-part segmentation at 25 fps on a 2018 15-inch MacBook Pro, and 21 fps on an iPhone X.” - Recognizing graphs from images
Do you remember the last time you’ve created a diagram on paper and found yourself re-creating it in a general diagram editor? - AI today and tomorrow is mostly about curve fitting, not intelligence
An “oldie” from 2018, though still relevant today. - Deep learning with Pytorch
Download a free copy of the Essential Excerpts from the book - How to recognize AI snake oil (presentation)
Much of what’s being sold as “AI” today is snake oil — it does not and cannot work. - Hacking Neural Networks
This is a fantastic collection of exercises if you want to learn to hack neural networks or their weaknesses in general!
