Many real-world problems require complex coordination between multiple agents — e.g., people or algorithms. A machine learning technique called multi-agent reinforcement learning (MARL) has shown success with respect to this, mainly in two-team games like Go, DOTA 2, StarCraft, hide-and-seek, and capture the flag. But the human world is far messier than games. That’s because humans face social…
As the pandemic wears on, humanoid robots that can perform general tasks could become increasingly important. They can move heavy objects in warehouses, assist with last-yard deliveries, and potentially be involved in other tasks where low human-to-human contact is…
Why AI can’t move forward without diversity, equity, and inclusion
November 26, 2020
The need to pursue racial justice is more urgent than ever, especially in the technology industry. The far-reaching scope and power of machine learning (ML) and artificial intelligence (AI) means that any gender and racial bias at the source is multiplied to the nth power in…
Google’s AI lets users search language-agnostic knowledge bases in their native tongue
November 26, 2020
Entity linking fulfills a key role in grounded language understanding. Given a text mention of an entity (e.g., the word “helpful”), an algorithm identifies the entity’s corresponding entry in a knowledge base (such as a Wikipedia article). To extend its usefulness, researchers at Google propose a new technique where language-specific mentions resolve to a language-agnostic knowledge base.
How Google built its AI-powered Hum to Search feature
November 26, 2020
In October, Google announced it would let users search for songs by simply humming or whistling melodies, initially in English on iOS and in more than 20 languages on Android. At the time, the search giant only hinted at how the new Hum to Search feature worked. But in a…
Researchers claim AI algorithm uses heart rate and motion data to predict age, sex, and more
November 26, 2020
How much information about a person’s behavior can you ascertain from a wearable device? Researchers at the University of Cambridge and Alan Turing Institute sought to investigate this in a study involving over 280,000 hours of wrist accelerometer and wearable…
Escooter startup Voi uses AI to detect pedestrians and sidewalks
November 26, 2020
European electric scooter startup Voi is turning to artificial intelligence (AI) to enable its vehicles to detect pedestrians and sidewalks, a move designed to ingratiate itself to municipalities worried about chaotic micromobility schemes.
Escooter (and ebike) sales have surged throughout 2020 as travelers sought alternatives to crowded public transport, but this proliferation serves to stoke…
Researchers at Google claim to have developed a machine learning model that can separate a sound source from noisy, single-channel audio based on only a short sample of the target source. In a paper, they say their SoundFilter system can be tuned to filter arbitrary sound…
AI research finds a ‘compute divide’ concentrates power and accelerates inequality in the era of deep learning
November 26, 2020
AI researchers from Virginia Tech and Western University have concluded that an unequal distribution of compute power in academia is furthering inequality in the era of deep learning. They also point to the impact on academia of people leaving prestigious universities for…
In addition to announcing the release of macOS Big Sur and formally revealing its first Mac-specific M1 chips, Apple today unveiled the first Mac computers that will use its new software and hardware: M1-powered versions of the 13-inch MacBook Air, 13-inch MacBook Pro, and Mac mini. Each of the machines will be using the M1 chip to deliver performance roughly three times the prior comparable Intel…