How-To -- Pure AI

How-To

Researchers Explore Differential Evolution Optimization for Machine Learning

Differential evolution may allow the creation of neural prediction systems that are more powerful than the current generation of systems.

By Pure AI Editors
08/02/2021

Researchers Use Machine Learning Techniques to Detect Compromised Network Accounts

Researchers and engineers at Microsoft have developed a successful system that detects compromised enterprise network user accounts.

By Pure AI Editors
07/07/2021

A lightbulb in front of a blue background

Researchers Explore Techniques to Reduce the Size of Deep Neural Networks

The goals are to save money (over $4 million for a single training run of a natural language model) and reduce CO2 emissions.

By Pure AI Editors
06/02/2021

Researchers Explore Intelligent Sampling of Huge ML Datasets to Reduce Costs and Maintain Model Fairness

Also, less CO2 is emitted, which is a good thing because one researcher said: "The current approach for building ML models is not sustainable and we will hit a ceiling soon, if not already."

By Pure AI Editors
05/03/2021

Researchers Devise a New Machine Learning Algorithm for Dataset Similarity

Calculating dataset similarity is difficult but is useful for several scenarios and can potentially save millions of dollars in ML computing costs and even greatly reduce carbon footprints.

By Pure AI Editors
04/08/2021

Researchers Explore Machine Learning Calibration

One of the main reasons for the increased interest in the tricky field of ML model calibration is the fact that the more complex a model is, the more likely the model is to not be well-calibrated.

By Pure AI Editors
03/03/2021

Warm-Start Training for Machine Learning

The challenge is that when new data arrives periodically, a new prediction model trained using the existing model (a "warm-start"), the resulting new model performs worse than a model trained from scratch (a "cold-start").

By Pure AI Editors
02/01/2021

Defending Machine Learning Image Classification Models from Attacks

By adding random noise to an image to be classified, and then removing the noise using a custom neural denoiser, standard image classification models are less likely to be successfully attacked.

By Pure AI Editors
01/05/2021

Computing the Similarity of Machine Learning Datasets

The Pure AI Editors explain two different approaches to solving the surprisingly difficult problem of computing the similarity -- or "distance" -- between two machine learning datasets, useful for prediction model training and more.

By Pure AI Editors
12/01/2020

Machine Learning Image Techniques for Biology Problems

"I think it's safe to say that the application of ML techniques to biology scenarios will bring new breakthroughs, some of which will be very surprising and unexpected," says a machine learning expert.