Facebook uses AI to rank posts

This is a review of some artificial intelligence news that happened throughout the week.

Processing the activity of 1.62 billion daily users, which generate 4 petabytes of data, along with 350 million photos is not easy. Facebook has 15,000 moderators to review everything from political content or harassment to terrorist threats and child exploitation. On Friday I announced the largest social network that will incorporate automatic learning for the process of moderation of information generated day by day. They will classify the issues that require a quick action and assign it to human moderators. Problematic posts will be evaluated on three criteria: Virality, severity, and likelihood that they are violating the rules.

Classifying Viruses with Machine Learning

A new system developed by researchers in Japan allows a single virion to be identified from common respiratory pathogens using a machine learning algorithm trained on changes in current through silicon nanopores. This research may prove useful for rapid and accurate detection tests for diseases such as COVID19 and influenza.

Artificial intelligence to forecast large-scale traffic conglomerations

Many cities are known for their traffic jams, in Los Angeles it is estimated that people spend 120 hours a year trapped in traffic. Based on this it is trying to solve by means of the fast prediction and potential to redirect the traffic. The idea is to apply machine learning after processing traffic pattern data collected from nearly a year from 11,160 sensors installed on California roads. It is intended that in milliseconds the model looks at the last hour of data and predicts the next hour of traffic with precision.

Nose with artificial intelligence to detect the freshness of the meat.

Scientists from Nanyang University of Technology, Singapore have created a system that mimics the nose of mammals to assess the freshness of meat. The nose has a bar code that changes color over time as it reacts to the gases produced by the meat as it decomposes. The algorithm based on a deep convolutional neural network predicts freshness with an accuracy of 98.5% compared to a commonly used algorithm that had an overall accuracy of 61.7%.

Thanks for reading, I hope you have informed yourself of something new. See you in the next edition.