Our blog
Leveraging Active Learning on Small Datasets
At Alectio, we’re usually known to help ML teams reduce the size of their datasets; our typical users come to us when they have more data than they know what to do with, and when training on their entire dataset is just not feasible, the reason being that most people...
Planes, Training and Anomalies (or How Much Data is Really Needed When Working with Sensor Data?)
If you are an avid Alectio follower, you’ll know by now that efficient training is what we do. It is not just a matter of philosophy (even though it is, as we have described here, here and here); it’s also a simple matter of economics. For many companies in a wide...
9 Reasons Why Active Learning is Not Widely Adopted
The Machine Learning field is full of buzz: deep learning, LSTMs, generative adversarial networks; the list keeps going on and on and on. Some of the most promising concepts, though, stand at the other end of the spectrum. Active learning has to be one of the top...
What computing forgetfulness on unseen data taught us
Neural networks do evolve and change their predictions as they train and there’s been a recent effort to use those changes to understand the underlying training data better. One of the more notable attempts at understanding these changes is an ICLR '19 paper by...
How Active Learning Can Massively Reduce Aerial Imagery Labeling Costs
As we enter 2021, active learning is perhaps the least understood and most underutilized technique in machine learning today. Its promise is simple and elegant: to reduce the overall records you use to train models without trading off accuracy. It’s an iterative,...
How Alectio Helped Voyage Dramatically Increase Their Performance and Development Speed with Active Learning
This article was written in collaboration with Voyage The field of computer vision reached a tipping point when the size and quality of available datasets finally met the needs of theoretical machine learning algorithms. The release of ImageNet, a fully-labeled...
Did Google Just Admit They Can’t Make AI Sustainable?
Perhaps you already know the story of Timnit Gebru, the high profile ethics researcher who was just forced out at Google, but if not, let’s level-set before we get started. Gebru is likely most famous for a paper she wrote while at IBM that highlighted the gender and...
There’s No Such Thing as Green AI
Meet John. John is a machine learning scientist. He likes his cat and he loves his dogs. He has a wife and two kids and he cares about the environment. He rides his bike to work instead of driving but when he has to drive, he takes his Tesla or his wife’s Prius. He...
5 Ways to Save on Data Labeling (Webinar Recording)
You probably don’t need us to tell you this, but machine learning is expensive. Whether it’s data storage, data warehousing, rising compute costs, models that need retraining, hiring the best and brightest in a competitive marketplace, or data labeling, spinning up...
Alectio Explains it All
If you work in machine learning, chances are you’ve had that moment at a party or a family dinner where someone asked you to explain some esoteric ML concept in a way your audience could understand. For experts, this can be tough. After all, machine learning is...
Increasing Accessibility to AI
The first cell phone ever sold was called the DynaTAC 8000X. The year was 1983 and for the low, low price of $3995, you could have been an early adopter. The phone itself took ten hours to charge and had a battery life of half an hour. It was a large, unwieldy thing...
Creating More Opportunities in AI
Last piece, we dug into the concept of increasing access to AI. In this one, we’re going to tackle opportunity. You may be asking yourself what exactly the difference is here, and, we’ll admit this a Venn diagram with some overlap. Diverse data labeling, for example,...
