Machine learning is often characterized as much an “art” as a “science” and in at least one regard, that’s true: its practitioners are prone to working under loosely controlled conditions, using training data that is being continuously tweaked with no versioning; modifying parameters during runs (because it takes too long to wait for the whole run before making changes); squashing bugs mid-run; these and other common practices mean that researchers often can’t replicate their own results — and virtually no one else can, either.
Pete Warden describes the frustration of machine learning researchers who can’t get their experiments to match published results or the models that emerge from them, and the first wave of responses to this, a movement to impose rigorous logging and note-taking on the research community.
In many real-world cases, the researcher won’t have made notes or remember exactly what she did, so even she won’t be able to reproduce the model. Even if she can, the frameworks the model code depend on can change over time, sometimes radically, so she’d need to also snapshot the whole system she was using to ensure that things work. I’ve found ML researchers to be incredibly generous with their time when I’ve contacted them for help reproducing model results, but it’s often months-long task even with assistance from the original author.
Why does this all matter? I’ve had several friends contact me about their struggles reproducing published models as baselines for their own papers. If they can’t get the same accuracy that the original authors did, how can they tell if their new approach is an improvement? It’s also clearly concerning to rely on models in production systems if you don’t have a way of rebuilding them to cope with changed requirements or platforms. At that point your model moves from being a high-interest credit card of technical debt to something more like what a loan-shark offers. It’s also stifling for research experimentation; since making changes to code or training data can be hard to roll back it’s a lot more risky to try different variations, just like coding without source control raises the cost of experimenting with changes.
The Machine Learning Reproducibility Crisis [Pete Warden]
(via 4 Short Links)
By design, data placed in the blockchain is visible to everyone in the world and can never be removed; everyone who mines bitcoin makes a copy of the blockchain, and so any illegal content stashed in the blockchain ends up on the computers of every miner.
A trio of engineering researchers from UC Berkeley modeled the effect of heavy reliance on GPS routing on municipal road efficiencies and found that people who are GPS-routed are likely to move to surface streets and secondary highways when the main highways are congested; though this increases overall throughput in a city and reduces overall […]
A team of material scientists from Northwestern University figured out how to make hair dye in various shades of grey, all the way to a very, very black black, out of graphene sheets.
The Nintendo Switch is king when it comes to gaming on the go, but it’s tough to lose yourself in Zelda: Breath of the Wild or Skyrim if your battery dies out. That’s where this Nintendo Switch Battery Charger Case comes into play. Built exclusively for Nintendo Switch, this pack allows for uninterrupted charging while you play, […]
Creative designers play a pivotal role in engaging target audiences and customers, and while companies are eager to bring more of these professionals on board, you’ll have a hard time getting your foot in the door if you’re not using the industry’s best tools. From Adobe to Maya, the eduCBA Design & Multimedia Lifetime Subscription Bundle […]
As more companies aim to reel in costs and boost productivity, project managers are becoming an essential part of many operations, and they’re paid handsomely for their expertise. But, while demand is high, you’ll have a hard time getting your foot in the door if you’re not toting the right certifications. The Official Lean Six Sigma […]